Skip to main content

Full text of "Aluminium; its history, occurrence, properties, metallurgy and applications, including its alloys"

See other formats

Hate ^allege of JVgricultutc 
At (Sfornell UniaerHitH 


Cornell University Library 
TN 775.R55 1896 

Aluminium; its history, occ"''* "SiPmSP^ 

3 1924 003 633 751 

Cornell University 

The original of tiiis book is in 
tine Cornell University Library. 

There are no known copyright restrictions in 
the United States on the use of the text. 












f . . 





No. 810 Walnut Stbebt. 



ST, dunstan's house, petteb lane, fleet stbeet, 



(5 13% S^ 




Feinted by the 


63 and 55 N. Queen Street, 

Lancastek, Pa., U. S. A. 



iFteberick Mobler 




Ibenti Sainte^Claite Seville 





Ten years ago aluminium was an almost unknown metal; 
then it sold for twelve dollars a pound, now it is bought for fifty- 
cents ; then the yearly production was less than is the present 
daily output ; then only three books had been written about it, 
since then seven have appeared and two journals have been es- 
tabHshed to represent it. The lowering of the price, the increased 
production, the wide circulation of reliable information about 
aluminium, have brought to pass the dream of Deville, for it has 
truly become a metal of every-day life. 

The abundance of aluminium in nature, the purity of its ores, 
its wonderful lightness and adaptability to numerous purposes, 
indicate that the goal of the aluminium industry will be reached 
only when this metal ranks next to iron in its usefulness to man- 

To assist this consummation by furnishing all the reliable in- 
formation about aluminium, to thus enlighten the general reader, 
help the workman, instruct the student, assist the experimenter, 
and in every way to speed the industry on its destined path, 
is the raison d'etre of this work. 

The writer thanks the public for its generous appreciation of 
his past endeavors. The volume has been revised to date, sev- 
eral chapters have been largely re-written, and every effort has 
been made to make the work worthy of the splendid industry 

which it represents. 

Joseph William Richards. 

The LEfflGH University, 
Bethlehem, Pa., October zj, iSgs- 



If it was true that no apology was necessary in presenting a 
work on Aluminium in English, as stated in the preface to the 
first edition of this book, it is equally true that still less apology 
is necessary in offering an improvement on that work. 

The present volume is designed to be an improvement on 
the former one in the following respects : Mistakes have been 
corrected wherever detected by the author or pointed out by 
his friends ; in some instances the order of treatment of different 
parts has been revised, so as to bring them into strict, logical 
sequence ; the more strictly historical processes are described in 
greater detail, in order to preserve a complete record of the rise 
of the aluminium industry; chapters have been added treating 
on the properties and the preparation of aluminium compounds, 
on the theoretical aspect of the reduction of aluminium com- 
pounds, and on the analysis of commercial, aluminium and its 
common alloys ; the original chapters have been in several cases 
sub-divided, and every part treated more by itself and in greater 
detail than before ; finally, additions have been made through- 
out recording and describing the progress achieved in the last 
three years, with a completeness which it is hoped is up to the 
standard of the rest of the book. 

The method of treatment in the present edition will be found 
to be more critical, for wherever a reasonable doubt might be 
expressed as to the correctness of certain claims, or a rational 
explanation advanced for certain phenomena, the author has not 



hesitated to put his best thought on the question and to state 
his conclusions unreservedly. 

The friendly criticisms of the scientific press and their sug- 
gestions have been kept in view in preparing this new edition. 
The spelHng " aluminium " has been retained, because no suffi- 
cient reasons have been advanced for changing it to " alu- 
minum;" and even if each way was equally old and as well- 
sanctioned by usage and analogy as the other, the author's 
choice would be the longer spelling, as being more euphonious 
and agreeable to the ear. 

It has been the author's endeavor to make this volume as 
complete as possible, as accurate as possible, to write it in a 
manner which will be entertaining to the general reader, and to 
furnish a treatise which will be of practical value to the practical 
metallurgist, as well as of scientific merit where it touches on 
matters of theory. 

J. W. R. 

Bethlehem, Pa., March 12, i8()o. 


No apology is necessary in presenting a work on Aluminium 
in English. In 1858 Tissier Bros, published in France a small 
book on the subject. H. St. Claire Deville, the originator of 
the aluminium industry, published a treatise, also in French, in 
1859. Deville's book is still the standard on the subject. Until 
December, 1885, we have an intermission, and then a work by 
Dr. Mierzinski, forming one of Hartleben's Chemisch-Technische 
Bibhothek, which is a fair presentation of the industry up to 
about 1883, this being a German contribution. Probably be- 
cause the English-speaking people have taken comparatively 
little hand in this subject, we find no systematic treatise on 
aluminium in our language. The present work aims to present 
the subject in its entirety to the English reader. 

Tissier, Deville, Mierzinski, and the German, French, and 
English scientific periodicals have been freely consulted and 
extracted from, full credit being given in each case to the author 
or journal. As this art has of late advanced so rapidly, it has 
been a special aim to give everything that has been printed up 
to the time of publication. 

The different parts of the work are arranged in what seemed 
their logical order, corresponding closely to that followed by 
Deville. The Appendix contains an account of laboratory ex- 
periments, etc., several of which, it is trusted, may be of value. 

In conclusion, the author wishes to thank the Faculty of his 
" Alma Mater," Lehigh University, for their permission to use 



his Thesis on Aluminium as the basis of this treatise ; also, to 
acknowledge his indebtedness to Dr. Wm. H. Greene, of Phil- 
adelphia, for assistance rendered in the preparation of the work 
for the press. 

J. W. R. 
Philadelphia, November 2j, j8S6. 


{Arranged chronologically.^ 

Tissier . Recherche de 1' Aluminium. C. & H. Tis- 

sier. Paris : J. Hetzel et Cie. 1858. 

Uhlenhuth Die Darstellung des Aluminiums. Ed. Uhl- 

enhuth. Quedlinburg: G. Basse. 1858. 

Deville De I'Aluminium. H. St. Claire Deville. 

Paris : Mallet-Bachelier. 1859. 

Mierzinski . Die Fabrikation des Aluminiums. Dr. Stan- 

islaus Mierzinski. Vienna : Hartleben's 
Chemisch-technische Bibliothek. 1885. 

Richards (ist Ed.) Aluminium. Joseph W. Richards. Phila- 
delphia : Henry Carey Baird & Co. 1887. 

Richards (2nd Ed.) Aluminium. Joseph W. Richards. Phila- 
delphia : Henr}' Carey Baird & Co. i8go. 

Le Verrier Note sur la Metallurgie de I'Aluminium. 

U. LeVerrier. Paris: BaudryetCie. 1891. 

Minet L' Aluminium. Adolphe Minet. Paris ; Ber- 
nard Tignol. 1892. 

Lejeal L'Aluminium. Adolphe Lejeal. Paris: J. 

B. Bailli^re et File. 1894. 


Fremy Enclyclop£die Chimique. Fremy. Paris : 

Ch. Dunod. 
Art : L'Aluminium par Margottet (1883). 
Art : L'Aluminium et ses Alliages par Wick- 
ersheimer (1890). 

Kerl and Stohman Enclyclopadisches Handbuch der Tech- 

nischen Chemie. 4th Ed. Brunswick : 
Vieweg und Sohn. 1886. 
Art: Aluminium von R. Biederman. 
Watts Watts' Dictionary of Chemistry. 

( xiii ) 



Ann. de Chim. et de Phys. . . Annales de Chimie et de Physique. 

Ann. der Chem. und Pharm. ) Liebig's Annalen der Chemie und Phar- 

Liebig's Ann J made. 

Bull, de la Soc. Chim Bulletin de la Soci^t^ Chimique de Paris. 

Chem. News The Chemical News. 

Chem. Zeit. . . Chemiker Zeitung (Cothen). 

Compt. Rend Comptes Rendus de les Stances de I'Acadd- 

mie. Paris, 

Dingl. Jrnl Dingler's Polytechnisches Journal. 

E. and M. J The Engineering and Mining Journal. 

Jahresb. der Chem Jahresbericht ueber die Fortschritte der 


Jrnl. Chem. Soc Journal of the Chemical Society. 

Jrnl. der Pharm Journal der Pharmacie. 

Jrnl. fiir pr. Chem Erdmann's Journal fiir praktische Chemie. 

Mon. Scientif. Le Moniteur Scientifique. Dr. Quesnesville. 

Phil, Mag The London and Edinburgh Philosophical 

Phil. Trans ... Transactions of the Royal Philosophical 


Pogg. Ann Poggendorff's Annalen. 

Poly. Centr. Blatt Polytechnisches Central-Blatt. 

Proc. Ac. Nat. Sci. .... Proceedings of the Academy of Natural 

Sciences (Philadelphia). 

Quarterly Journal Quarterly Journal of the Society of Arts. 

Rpt. Brit. A. A. S Report of the British Association for the 

Advancement of Science. 

Sci. Am. (Suppl.) Scientific American (Supplement). 

Wagner's Jahresb Wagner's Jahresbericht der Chemischen 

Zeit. fiir anal. Chem Zeitschrift fiir Analytische Chemie. 


The Aluminum World .... New York. Published monthly. Estab- 
lished September, 1894. 

L'Aluminium Paris. Published monthly. Established 

January, 1895. 



History op Ai<uminium. 


The alumen of Pliny; Alumen Roccae or rock alum; Alums and vitriols 
confounded .... ....... 1 

Paracelsus distinguishes alums from vitriols; Views of Ettmiiller, Stahl, 
Hoffman, Geoffroy, Hellot and Pott, on the nature of the base of 
alum; Marggraff's dissertations on alum and its earth; Morveau fixes 
the nomenclature of alumina 2 

The search for aluminium; Macquer's supposition; Baron's attempts to 
reduce alumina; Lavoisier's views on the nature of alumina . . 3 

Ruprecht and Tondi's experiments; Criticisms of Savaresi, Klaproth 
and Karsten; Dr. Hare melts alumina without noticing any reduction; 
Sir Humphry Davy attacks the earths with his great battery . . 4 

Davy's experiments on reducing alumina; He reduces it in the electric 
arc in presence of iron; He suggests a name for the metal ... 5 

Prof. Benjamin Silliman's experiment with Dr. Hare's blow-pipe; Oer- 
sted's attempt to reduce aluminium chloride; Berzelius' attempt to 
reduce cryolite 6 

Professor Frederick Wohler, of Gottingen, isolates aluminium as a 
metallic powder, in 1827 7 

Wohler obtains the metal in globules, in 1845; Professor Henri Saint- 
Claire Deville obtains aluminium en masse, in 1854 . ... .8 

Deville's first attempts; His paper read before the French Academy on 
February 6, 1854 . . 9 

The Academy aids Deville in his further work ; A reclamation by 
Chenot; Further experiments at the Ec61e Normale; Reduction of 
aluminium chloride by the battery 10 

Deville's experiments in manufacturing sodium; Installation of the 
sodium process at Rousseau Bros, chemical works at La Glaci^re; 
Deville and Bunsen publish their electrolytic methods; The Emperor 
Napoleon III. becomes interested and defrays the expense of further 
experiments H 




TheTissier Bros, experiment on producing sodium; Deville's work at 
Javel; The industry on a firm basis ' ^^ 

Exhibit of aluminium at the Paris Exposition, 1855; Misunderstanding 
between Deville and the Tissier Bros 13 

Tissier Bros, start an aluminium works near Rouen; History of this 

enterprise ^^ 

Deville and his friends put up a large plant at La Glaci&re; Removal to 
Nanterre; Removal to Pechiney and Co.'s works at Salindres; Tissier 
Bros, publish their "Recherchessurl'Aluminium" {1858) . . 15 

Deville publishes his "De I'Aluminium" (1859); Hi^ conclusions as to 
the usefulness of aluminium 16 

Experiments of Dr. Percy and H. Rose on reducing cryolite; Manufac- 
ture of aluminium by A. Monnier, of Camden, N. J.; Calculation of 
cost of the new metal by W. J. Taylor 17 

The first aluminium works in England, at Battersea (1858); Bell Bros, 
works near Newcastle-on-Tyne; Project to start works at Berlin; Dr. 
Clemens Winckler's retrospect, 1879 18 

Condition of the aluminium industry in 1879 19 

The industry from 1879 to 1882; W. Weldon's views on its prospects in 
1883 . 20 

The search for a substitute for sodium ; Webster's improvements in the 
industry .21 

H. Y. Castner's invention for producing sodium; His experiments in 
New York 22 

Castner's demonstration in England; Unites with Webster to form the 
"Aluminium Company, Limited; " Large works at Oldbury near Bir- 
mingham, stop making aluminium in 1891 ...... 28 

Revival of electric methods, using dynamos instead of the battery; 
Gratzel's process 24 

Kleiner's process; Hall's process, invented in 1886, essentially different 
in principle from previous attempts 25 

Incorporation of the Pittsburgh Reduction Co. to work Hall's process; 
Works at Kensington, Pa., and Niagara Falls; Minet's experiments 
at Creil and works at Saint Michel, Savoy 26 

The electric-furnace methods; Siemens furnace; Grabau's experiments; 
Mierzinski's remarks 27 

Cowles Bros, process, 1885; Plant at Lockport, N. Y., and works in 
England; The principle of the process . 28 

H^roult's processes; Principle of his alloy process 29 

The "Aluminium Industrie Actien Gesellschaft " formed to work 
HSroult's processes; Dr. Kiliani, manager of their works at the Falls 
of the Rhine; The Hiroult process in France 30 

Experiments with H^roult's process in the United States; The "Alliance 
Aluminium Co., of London," formed to work Netto's sodium process. 31 

Failure of the Alliance Aluminium Co.; Grabau's improvements in pro- 



ducing sodium and aluminium fluoride; Colonel Frishmuth's works 
in Philadelphia 32 

Exhibit of aluminium at the Paris Exposition, 1889; Progress of the in- 
dustry since 1890 33 

The Cowles Electric Smelting Co. sued for infringement of the Hall 
process; Exhibits of aluminium at the Columbian Exposition in Chi- 
cago, 1893; The production of aluminium one of the great achieve- 
ments of the century 34 

Statistics; The decline in the price of aluminium and aluminium bronze. 35 

Production of aluminium in France from 1854 to 1892; Production in 
England; In Switzerland 36 

Production of the United States from 1883 to 1894; Total aluminium 
produced in the world to the end of 1892; World's production in 1893; 
An estimate of the production in 1894 37 

The probable production in 1895 38 


Occurrence of Aluminium in Nature. 

Clarke's estimate of the proportion of aluminium in the earth's crust; 
Not found metallic; Most frequent combinations in rocks, soils and 
clays; Occurrence in some plants; Ricciardi's experiments on the 
assimilation of alumina by plants 39 

Composition of the most frequent aluminium minerals; Bauxite the 
most useful mineral for the aluminium industry 40 

Localities in which bauxite occurs ; Deposits worked in the United 
States; Value per ton 41 

Analyses of foreign bauxites 42 

Analyses of American bauxites 43 

Laur's generalizations concerning French bauxites . . . .44 

Hunt's generalizations on American bauxites; The occurrence of cryo- 
lite in Greenland 45 

Composition of cryolite; Amount imported into the United States; De- 
scription of a small deposit in Colorado 46 

Analysis of the Colorado cryolite; The sources of corundum . . 47 

Mining of corundum in the United States; The composition and proper- 
ties of kaolin 48 

Localities of the large kaolin beds; Value of kaolin as an aluminium 
ore; Common clays 49 

Native sulphate of alumina; Deposits in New Mexico, Colorado and 
Ohio 50 

Economic value of the aluminium sulphate deposits . . . .51 



Physicai, Properties of Ai,uminium. 


Commercial aluminium not chemically pure; The most frequent im- 
purities 52 

The influence of small amounts of impurities on the physical properties 

of the metal 53 

Analyses of commercial aluminium ; Notes on the analyses . . .54 

Recent analyses of present commercial metal 55 

Difference in the composition of metal made by the sodium and the 
electrolytic methods; The different conditions in which silicon exists 
in aluminium; Rammelsberg's analyses ...... 56 

Absorption of gases by molten aluminium; Dumas' analyses; Author's 
observations; Tie Verrier's detection of carbon in commercial alumin- 
ium . ..... 57 

Influence of carbon on the physical properties of aluminium; Le Ver- 
rier detects nitrogen in commercial aluminium ... .58 

Influence of nitrogen on the physical properties of aluminium; The 
color of pure and impure aluminium ....... 59 

Tarnishing of aluminium in the air; Fracture of purest aluminium . 60 
Fracture of impure ahiminium; Hardness; Testing quality -with the 

knife; Hardening by cold working .61 

Comparison of the hardness of aluminium with other metals; Specific 

gravity . 62 

Influence of impurities on the specific gravity; Increase of specific 
gravity by working . . ....... 63 

Comparison of the specific gravity of aluminium and other metals . 64 
Comparative value of equal weights and equal volumes of the different 

metals; Fusibility 65 

Melting point determined by the electric pyrometer; Influence of im- 
purities on the melting point; Volatilization; Odor .... 66 

Taste; Magnetism; Sonorousness 67 

Sound produced by an aluminium bell, bar and tuning-fork; Alumin- 
ium sounding boards 68 

Velocity of sound in aluminium; Crystalline form; Elasticity . . 69 
Tensile and compressive strength ........ 70 

Table of tensile tests . . 71 

Table of compressive tests; Transverse tests . . . . 72 

General conclusions of Messrs. Hunt, Langley and Hall; Variation of 

tensile strength with the temperature 73 

Influence of working and annealing on the mechanical properties; 

Tensile strength in comparison with weight 74 

Comparison of the tensile strength with that of other metals, section 
for section and weight for weight; Malleability . ... 75 



Thickness of aluminium leaf; Color by transmitted light; Ductility . 76 

Expansion by heat; Specific heat 77 

Determinations of the specific heat by Regnault, Kopp, Mallet, Tom- 
linson, Naccari, Le Verrier 78 

Specific heat and latent heat of fusion as determined by Pionchon and 
the author 79 

The amount of heat in molten aluminium; Agreement with Dulong 
and Petit's law 80 

Slow melting of aluminium due to its great latent heat of fusion; Com- 
parison with other metals; Electric conductivity ..... 81 

Professor Dewar's determinations of the electric conductivity at very 
low temperatures; Conductivity for heat ...... 82 


Chemical Properties of Aluminium. 

Atomic weight; Action of air . . . . , . . . .83 
Resistance to air when melted; Action of air and water together; 

Piqures 84 

Combustion of aluminium powder or leaf ; Action of water . . .85 
Decomposition of water by aluminium leaf; Action of hydrogen sul- 
phide and sulphur 86 

Experiment of the author with hydrogen sulphide and molten alu- 
minium . 87 

Action of sulphuric acid; Ditte's explanations 88 

Le Roy's quantitative tests with sulphuric acid 89 

Action of nitric acid; Ditte's observations 90 

Experiments of Le Roy, Lunge and the author on the action of nitric 
acid; Use of aluminium in the Grove battery; Action of hydrochloric 

acid ... .91 

Action of hydrobromic, hydriodic and hydrofluoric acids; Experiment 

by the author; Action of organic acids, vinegar, etc 92 

Action of the acids found in food; Deville's prediction of the usefulness 
of aluminium for culinary utensils; Lubbert and Roscher's experi- 
ments with thin foil 93 

Balland's exhaustive tests with foods ; Rupp's results; Lunge and 

Schmid's experiments 94 

Details of Lunge and Schmid's tests 95 

Conclusions from these tests; Confirmation of these results by actual 

experience of two years with aluminium culinary utensils . . -96 
Action of common salt, molten and in solution; Sea- water; Aluminium 
boats ■ ^'^ 



Action of organic secretions, the saliva, etc 98 

Action of caustic potash, caustic soda; Quantitative tests . . .99 

Action of ammonia; Lime water; Solutions of metallic salts . . . 100 
Action of solutions of mercury, copper, silver, lead .... 101 

Action of solutions of zinc, alkaline chlorides, aluminium salts; Action 

of melted fluorspar; Use as a flux 102 

Use of cryolite as a flux; Action of molten silicates and borates . . 103 
Action of fused nitre; Purification of impure aluminium by nitre; Re- 
action with alkaline sulphates and carbonates 104 

Reaction of aluminium on oxides of manganese, zinc, iron, lead, copper, 

barium 105 

Greene and Wahl's process of producing pure manganese; Action of 
phosphate of lime, silver chloride, mercurous chloride, carbonic oxide, 
hydrogen, chlorine, bromine, iodine, fluorine 106 


Properties and Preparation of Aluminium Compounds. 

General considerations; Position of aluminium in the periodic classifi- 
cation of the elements . 107 

Observations on the structure of aluminium compounds . . . 108 

The tri-valency of aluminium ; Aluminates ...... 109 

Neutral salts of aluminium; Basic salts; General methods of formation 

of aluminium salts ... 110 

General properties of aluminium salts; Chemical reactions in solution; 

Reactions before the blowpipe Ill 

Alumiuium oxide; Physical and chemical properties . . . .112 
Hydrous aluminium oxides, natural and artificial; The soluble hydrate. 113 
Aluminates; Potassium aluminate; Sodium aluminates; Barium alumi- 

nate ...... 114 

Aluminates of calcium, zinc, copper, magnesium, iron, beryllium; Alu- 
minium chloride 115 

Hydrous aluminium chloride; aluminium-sodium chloride . . .116 
Double chlorides of aluminium with phosphorus, sulphur, selenium, 

ammonium 117 

Aluminium-chlor-sulphydride ; Aluminium-chlor-phosphydride ; Alu- 
minium bromide; Aluminium iodide 118 

Aluminium fluoride; Methods of production 119 

Fluorhydrate of aluminium; Aluminium-hydrogen fluoride; Alumin- 
ium-sodium fluoride .120 

Aluminium sulphide; Double sulphides with sodium and potassium . 121 
Use of aluminium sulphide in the laboratory for generating hydrogen 
sulphide; Aluminium selenide; Aluminium borides . . . .122 



Aluminium boro-carbides; Aluminium carbide 123 

Production of aluminium carbide in the electric furnace by Le Verrier . 124 
Composition and properties of aluminium carbide; Aluminium nitride 125 

Aluminium sulphates, anhydrous and hydrous 126 

Basic aluminium sulphates 127 

Alums; Potash alum; Ammonia alum .... . . 128 

Soda alum; Bouble sulphates of aluminium and the metals; Aluminium 

selenites 129 

Aluminium nitrate; Aluminium antimonate; Aluminium phosphates . 130 

Aluminium carbonate; Aluminium borate 131 

Hydrous aluminium borates; Aluminium silicates; Blast furnace slags 132 


Preparation of Aluminium Compounds for Reduction. 

The preparation of alumina; Preparation from alums or aluminium sul- 
phate .... .133 

Preparation of alumina from ammonia alum ...... 134 

Method used by Deville at Javel; Tilghman's method . . . 135 

Webster's process of making alumina from potash alum . . 136 

Preparation of alumina from bauxite ; Method uSed at Salindres . . 137 
Heating with soda and washing out the aluminate ... . 138 

Apparatus used for washing; Precipitation of alumina by carbonic acid 

gas 139 

Apparatus used for the precipitation ; The barattes . . . 140 

Washing and drying the precipitated alumina; Behnke's method of 

treating bauxite; Lieberfs method 141 

Muller's method for extracting alumina from silicates; Use of common 

salt to decompose bauxite; Wagner's method of treating bauxite . 142 
Laur's method of using sodium sulphate to decompose bauxite; Lowig's 
experiments on precipitating alumina from sodium aluminate solu- 
tion; Dr. Bayer's improvements . 143 

Preparation of alumina from cryolite; The dry way .... 144 

Furnace for heating cryolite with chalk; Thomson's method . . 145 

Conduct of the operation ; Washing the fusion .... 146 

Production of the carbonic acid gas for precipitation .... 147 
Composition of the precipitate; Action of steam on molten cryolite; 

Utilization of aluminous fluoride slags 148 

Utilization of the residual slags from sodium retorts by fusing with the 

aluminous slags; Decomposition of cryolite in the wet way . . 149 

Method used by Deville at Javel; Modification of this process by Sauer- 
wein ............. 150 

Reactions in Sauerwein's process; Weber's method of decomposing 
cryolite; Schuch's reaction 151 


Preparation of aluminium chloride and aluminium-sodium chloride . 152 
Method used by Beville on a small scale at Javel . . . 1 53 

Manufacture by Deville on a large scale 154 

Furnace used and working of the operation . . . 155 

Output of the furnace; Purification of the crude chloride . . .156 
Deville's idea of making the double chloride; Margottet's description 
of the apparatus used at Salindres . ... 157 

Conduct of the operation at Salindres . . . . . 158 

Cost of making the double chloride at Salindres . . . 159 

Plant of the Aluminium Company, Limited, at Oldbury, near Birming- 
ham, England . . . . _. . . 160 

Arrangements of the furnaces and conduct of the operation . . .161 
Output of this plant; Purity of the double chloride . .162 

Castner's method of purifying the double chloride . ... 163 

Cost of the purified double chloride; Gadsden's method of producing 

aluminium chloride; Count R. de Montgelas' improvement . . 164 
Prof. Mabery's process for producing aluminium chloride in the electric 
furnace; Curie's method; Warren's general process for anhydrous 
metallic chlorides . . ... 165 

Camille A. Paure's process; Description by Berthelot . 166 

Details of Faure's process; Output claimed; DuUo's method of making 
aluminium chloride from clay ... ... 167 

Remarks on Dullo's suggestions; The preparation of aluminium fluoride 
and artificial cryolite . ....... 168 

Berzelius' method of making artificial cryolite; Deville's methods . 169 
Pieper's reactions in the wet way; Bruner, Deville and Hautefeuille on 
making aluminium fluoride . . . . 170 

Grabau's processes; Aluminium fluoride from aluminium sulphate and 
fluorspar . . ... ... 171 

Aluminium fluoride indirectly and directly from kaolin . . . 172 

Precautions necessary to obtain it free from iron 173 

The preparation of aluminium sulphide; Fremy's researches . . . 174 
Purity of the aluminium sulphide produced .... . 175 

Reichel's results; Experiments of the author . .... 176 

Propositions of Comenge, Petitjean, etc. . . • 177 

Lauterborn's patent anticipated . .... 178 


The Manufacture of Sodium. 

Relations of the production of sodium to the aluminium industry; Isola- 
tion of sodium by Davy, Gay Lussac and Thenard, Curaudau . . 179 

Briinner's apparatus; Donny and Mareska's condenser; Deville's im- 
provements at Javel (1855) . ]80 





CONTENTS. xxiii 


The chemical and physical properties of sodium . ... 181 

Method of reduction used at Javel; Composition of the mixtures for re- 
duction .... .... ... 182 

Utility of the different ingredients of the mixtures; Difficulties of con 

densing sodium vapor . . 

Mixture used at La Glacifere and Nanterre; Charging the apparatus 
Description of retorts and condensers ; Manufacture in mercury 
bottles ... . . 

The furnace used for manufacture in mercury bottles 

Detailed description of condensers 

Conduct of the operation . . ... 

Handling of the condensed sodium; The temperature necessary for re 
ductiou ... .... 

Use of decarburized, cast-iron bottles; The continuous manufacture in 

Furnace for manufacture in cylinders 

Arrangement of the cylinders in the furnace 192 

Charging and discharging the cylinders . .... 193 

Tissier Bros. ' method of procedure at Rouen 194 

Furnace used by Tissier Bros ... 195 

Charging and discharging Tissier's furnace 196 

Causes of loss of sodium; Deville's improvements at La GlaciSre (1857) 197 
Deville's large cylinder furnace; Difficulties in its operation . . 198 

Deville's attempts to use cast-iron retorts; Deville's improvements at 
Nanterre (1859) ... . . ... 199 

Necessity of protecting the iron cylinders; Deville returns to smaller 

apparatus ... . 200 

Cost of sodium at Salindres in 1872; An experiment of Deville's in 1864; 
Wagner's method of preserving sodium .... . 201 

Reduction of potassium and sodium compounds together; Thompson 
and White's method . ... ... 202 

Blackmore's process of reducing carbonate of soda; Thowless' ap- 
paratus; Jarvis' device for using fire-clay apparatus; Castner's process 203 

History and principle of Castner's invention 204 

Reaction on which the process is based; Form of furnace used . 206 

Experimental furnace erected in England 

Conduct of the operation in this furnace . 

Yield of the furnace; Composition of the residues . 

Cost of sodium by the Castner process 

Erection of large works to operate the Castner process 

Description of the largest furnaces .... 

Manner of working and output of these furnaces 

Preservation of the sodium; Reactions in the process 

Netto's sodium process; The Alliance Aluminium Company 

Details of Netto's apparatus 




Reduction of sodium compounds b3' electricity; Jablochaff's apparatus; 

Roger's method 219 

Details of Roger's experiments ... 220 

Grabau's observations on the electrolysis of molten sodium chloride . 222 
Details of Grabau's electrolytic process; Castner's process of electrolyz- 

ing caustic soda .... 223 


Thb Reduction of Aluminium Compounds from the Standpoint of 
Thermai, Chemistry. 

The proper way to use the data of thermo-chemistry .... 226 

Critical temperatures below which reactions do not occur . . . 227 
Influence of physical state in determining a reaction; The heat of forma- 
tion of some aluminium compounds ...... 228 

The heat of formation of alumina compared with that of other oxides; 
Elements which reduce alumina easily ..... 229 

Possibilities of the reduction of alumina ....... 230 

Reduction of alumina by carbon; Necessary conditions to be fulfilled . 231 
Heats of formation of chloride, bromide and iodide of aluminium, and 

other metals 232 

Deductions as to the agents capable of reducing these aluminium salts 233 
Heat of formation of aluminium fluoride and other metallic fluorides . 234 
Observations on the reduction of aluminium fluoride; Heat of formation 

of aluminium sulphide and other metallic sulphides . . . 235 

Observations on the reduction of aluminium sulphide; Electricity as a 

reducing agent ... 236 

The voltage necessary to produce decomposition of metallic fluorides, 
chlorides, oxides and sulphides; Tabular view of these data; Observa- 
tions thereon ... ...... 237 

Electrolysis of baths containing several ingredients; Illustration from 

Hall's process ... . 938 

Illustration from Minet's process; The voltage required to decompose 

alumina at different temperatures . 239 

Calculation of the temperature at which hydrogen can begin to reduce 

alumina; The same calculation for carbon . . . 240 

Alumina has been reduced by carbon in an iron blast furnace; Peculiar- 
ity of acetylene gas as a reducing agent . . ... 241 
Thermal aspect of other reactions used in the aluminium industry; The 

production of aluminium chloride from alumina ..... 242 
The production of aluminium sulphide from alumina; Peculiarity of 
carbon bi-sulphide from a thermal standpoint • . . . . 244 



Reduction op Aluminium Compounds by Means of Potassium or 



Classification of methods ; The reduction of chlorine compounds ; 

Oersted's experiments (1824) 246 

Wohler's experiments (1827); Wohler was repeating Oersted's work. 247 
Wohler isolates aluminium as a metallic powder ; His experiments 

in 1845 . 248 

Globules of aluminium obtained by Wohler in 1845 . . . 249 

Impurity of Wohler's metal; Deville's experiments in 1854 . 250 

Deville obtains pure aluminium in ingots ...... 251 

Deville's methods devised in his laboratory and applied at Javel (1855). 252 

Details of the apparatus employed; Conduct of the operation 253 

Deville's dissatisfaction with this apparatus and its results . . 255 

Deville reduces aluminium chloride by sodium vapor . . . 256 

Deville's perfected process as used at Nanterre (1859) ■ • • ^^'^ 

Details of Deville's improvements . . 258 

Reduction in crucibles; Lining of crucibles ...... 259 

Reduction on the bed of a reverberatory furnace . ... 261 

Deville's views on the aluminium industry in 1859 . . 263 

The Deville process as operated at Salindres in 1882 . . . 264 

The chemical reactions in the process ... . 265 

Description of furnace used and mode of operation . . . 266 

Cost of aluminium at Salindres; Niewerth's process . . 268 

Gadsden's sodium vapor process; Frishmuth's claims . . . 269 

H. von Groussillier's improvement; The Deville-Castner process . 270 

Description of Castner's works; Conduct of the operations . . . 271 

Efficiency of the process; Closing of these works ... . 273 


REDUCTION OF Aluminium Compounds by Means of Potassium or 
Sodium {Continued). 

Methods based on the reduction of cryolite; Experiments of H. Rose 

(1855) .... 274 

Experiments of Percy and Dick (1855) 

Deville's methods (1856-8) . . . 

Tissier Bros', methods; Impurity of the metal 
Wohler's improvements on Tissier's method; Gerhard's furnace 
Thompson and White's patent; Hampe's experiment; Netto's experi- 
ments at Krupp's works at Essen ... .... 290 





Formation of the Alliance Aluminium Co. to work Netto's process . 291 
Description of Netto's processes as actually carried out .... 292 
Return from Netto's process; The apparatus used at Essen . . 293 

Netto's other devices; Failure of the Alliance Aluminium Co. ; Grabau's 

process (1887) 294 

Advantages of Grabau's method of reducing aluminium fluoride . 295 

Reactions in Grabau's process; Description of the operation . . 296 

Description of the furnace and apparatus used ..... 297 

Great purity of the aluminium produced by Grabau .... 299 


Reduction of Aluminium Compounds by the Use of Ei<ectricity. 

Preliminary observations on the laws of electrolysis . . . 301 

Calculation of the voltage necessary to decompose aluminium com- 
pounds . . . . . . . . 302 

Utilization of such calculations; Electrolysis of aqueous solutions 303 

Electrolysis using a soluble anode .... ... 304 

Electrolysis using insoluble anodes . 305 

Electrolysis of baths containing several compounds .... 306 
Deposition of aluminium from aqueous solution; Thomas and Tilly's 

patent . . 307 

Corbelli's process; Thompson's claims; Gore's experiments . . . 308 

Dr. Gore's error; Jeangon's process 309 

Successful plating by the Harvey Filley Plating Co., of Philadelphia; 
Bertrand's assertion; Braun's methods ....... 310 

Dr. Fischer's experience with Braun's method; Farmer's apparatus; 
Senet's process ; Frishmuth's plating ; Overbeck and Niewerth's 
patent . . ... .311 

Reinbold's recipe; Count R. de Montgelas' patents . 312 

A. Walker's methods of procedure . ... . . 313 

Bull's proposition; Burghardt and Twining's patents .... 314 

Patents of Gerhard and Smith; Taylor and Coulson; Experience of 

Sprague, Dr. Winckler and Dr. Gore 3I5 

Opinions of Dr. Mierzinski, Hampe and Watt; Experiments of Dr. 

Lisle; Aluminium plating on the Philadelphia Public Buildings . 316 

Aluminium plating by Mr. Darling at the Tacony Metal Co. 's works, 

Philadelphia gjy 

Non-aqueous electric processes; Difficulty of a satisfactory classification. 318 
Davy's experiments (1810); Duvivier's experiment (1854) . . .319 
Bunsen's and Deville's methods (1854) ... . . 320 

Details of Deville's apparatus 32i 

Details of Bunsen's apparatus . . .... 322 


Deville and Caron's method of plating aluminium on copper . . 323 

Deville's experiments not an anticipation of Hall's process; Le Chatel- 

lier's patent . ... 324 

Monckton's proposition; Gaudin's process; Kagensbusch's process . 325 
Berthaut's proposition to use dynamos; Gratzel's process . . . 326 
Apparatus used by Gratzel; Later modification of his process . . 327 

Prof. Fischer's criticism of Gratzel's process; Operation of the process 
at Bremen . . .... .... 328 

Abandonment of the Gratzel process; Cowles Bros.' process . . . 329 
Professor Mabery's description of the Cowles electric furnace . . 330 

Dr. T. Sterry Hunt's remarks ... 331 

Bradley and Crocker's form of retort used by the Cowles Co. . . 382 

A. H. Cowles' form of furnace for continuous working; W. P. Thomp- 
son's paper on the Cowles process, read in England .... 333 
Cowles Bros', plant at Lockport, N. Y.; Details of the furnaces . . 334 

Charging the furnaces . 335 

Details of the operation of Cowles' furnaces 336 

The enlarged plant at Lockport . 339 

The Cowles Syndicate Co.'s works at Stoke-on-Trent, England . . 340 

Analyses of Cowles Bros', aluminium bronze 341 

Analyses of ferro-aluminium and slags . 342 

The reactions in Cowles Bros', process 343 

Electrolysis necessarily a secondary factor in the operation of these 

furnaces .... 344 

H. T. Dagger's experiment with the alternating current in a Cowles' 
furnace ....... .... 345 

History of the suit — Pittsburgh Reduction Co. vs. Cowles Electric 

Smelting and Aluminium Co. 346 

Menges' patent; Farmer's process; Kleiner's process . . 347 

Kleiner's experiments at Schafthauseu ; Experimental plant in 

England 348 

Dr. Gore's report on Kleiner's process 349 

Output of aluminium by this process; Purity of the metal . . . 352 
Kleiner's process outclassed by later processes; Lossier's proposition to 

decompose natural silicates of aluminium 
Omholt's furnace; Minet's process . 
Apparatus used in Minet's process . 
Composition and properties of the baths . 
Minet's views on the reactions in his baths 
Electrical measurements on Minet's pots 
Separation of iron and silicon by a carefully regulated current . . 360 
Output and quality of aluminium; Installations of this process . . 361 
Minet's plant now working on the Hall process; Feldman's process . 362 
Uselessness of Feldman's methods; Warren's experiments . . .363 
Zdziarski's patent; Grabau's apparatus 364 


xxviii CONTENTS. 


Roger's process; Details of experiments ....... 365 

Dr. Hampe on the electrolysis of cryolite . . ... 368 

History of Hall's experiments; Interference with H^roult in the United 

States patent ofi&ce . ... .... 373 

Specifications and claims of Hall's patents 374 

Installation of Hall's process at Pittsburgh 376 

Conduct of the operation in Hall's pots 377 

Consumption of carbon anodes 378 

Alumina used in the process; Efficiency of the electrolysis . . . 379 
Estimate of cost of aluminium by the Hall process in 1889; Enlarge- 
ment of the plant in 1890 381 

Removal and enlargement of the plant in 1891 ; The Niagara Falls 

plant 382 

Total output of the Pittsburgh Reduction Co.'s works; Efficiency of 

the process . 383 

Chemical reactions in Hall's baths; Alumina is the compound primarily 

decomposed ... ...... . 384 

Theory of an aluminium oxy-fluoride not proven; Hdroult's processes. 386 

Similarity of Hferoult's first process to that of Hall .... 387 

HSroult's alloy process; Installation at the Falls of the Rhine . . 388 

Description of furnaces used . ....... 389 

Conduct of the operation ... 390 

Output at full efficiency; Theoretical output ..... 392 

H^roult's alloy process not essentially electrolytic in its operation . 393 

HSroult's apparatus for pure aluminium .... 394 
Recent enlargements at the Rhine Falls ; The Heroult process in 

France 395 

Hdroult's alloy process at Boonton, N. J 896 

Details of cost of pure aluminium atNeuhausen; Faure's electrolytic 

process ... 398 

Winkler's patent; Willson's crucible 399 

Grabau's electrolytic process; Bucherer's patent ..... 400 

The Neuhausen sulphide process . 401 



Reduction by carbon without the preseuce of other metals; Chapelle's 

attempt ... 402 

Reinar's experiment with alum; Thowless' claim . ... 403 

Patent of Pearson, Liddon and Pratt; Bessemer's ideas .... 404 
Reduction by carbon and carbon dioxide; Morris's claims . . . 405 



Ideas of P. A. Emanuel; Reduction by hydrogen 406 

Gerhard's #patent; Success of H. Warren in reducing alumina by 

hydrogen 407 

Reduction by carburetted hydrogen; Fleury's experiments; Petitjean's 

statement 408 

Experiments of L^b^deef, Dr. Lisle, R. E. Green 409 

Reduction by cyanogen; Knowles' patent; Corbelli's process; Experi- 
ments of Deville and Lowthian Bell 410 

Reduction by double reaction; Comenge's process; Niewerth's furnace 411 
Process of G. A. Faure; Reduction by or in presence of copper; Calvert 

and Johnson's experiments ........ 413 

Evrard's method of making aluminium bronze; Benzon's patent . 414 

Faure's patent .... 415 

Experiments by BoUey, List, Dr. Lisle and Hampe . . .416 

Experiments of Reichel and the author; Processes of A. Mann and L. Q. 

Brin ... . . .... 417 

Reduction by or in presence of iron; Experiments of Comenge, Lauter- 

born, Reichel and the author; Niewerth's process .... 418 

Thompson's method 419 

Calvert and Johnson's experiments 420 

Chenot's reclamation; Use of spongy iron . . . . 421 

Faraday and Stodart's investigations; Wootz steel 422 

Presence of aluminium in pig-iron ; Reports by various analysts . . 423 
Billings' attempt to reduce alumina in contact with iron; Cleaver's 

patent . . ... . . 424 

Ostberg's statement about aluminium in pig-iron; Brin Bros.' process . 425 
Baldwin's process for treating foundry iron . . . . 426 

Alloy of the Williams Aluminium Company; Bamberg's method . . 427 
Aluminium in pig-iron from Marshall Bros.' Newport furnace . , 428 
Reduction by or in presence of zinc; Statements of Bdk^toff, DuUo, 

Basset . ... 429 

Wedding's remarks on Basset's process; Experiment by the author . 431 
Works of F. J. Seymour at Findlay, Ohio ... . . 432 

Lauterborn's method of reduction by zinc . .... 433 

Clark's doubtful processes .... ... 434 

Dr. Lisle on reducing aluminium sulphide by zinc; Bamberg's patent; 

Reduction by lead ... 435 

Reduction by manganese; Weldon's claim; Greene and Wahl's observa- 
tions t 436 

Reduction by magnesium; Gratzel's patent; Roussin's statement; Mont- 

gelas' patent; Winkler's experiments on alumina . . . 437 

Reduction by antimony; Lauterborn's formulas; Experiment by the 

author 438 

Reduction by tin; Process of Howard and Hill; Dr. Lisle's successful 

experiments . 439 



Reduction by phosphorus; Grabau's process . . ... 440 
Reduction by silicon; Wanner's claims «• -441 


Working in AItItminium. 

Melting; Deville's recommendations 442 

Biederman's directions; The author's methods . ... 443 

Bauxite and magnesite linings for crucibles and furnaces . . 444 

Casting; Deville's remarks 445 

Casting under pressure . ..... . 446 

Aluminium hollow- ware casting; Purification of aluminium; Freeing 

from slag . . 447 

Freeing from dissolved impurities 449 

Removal of zinc by distillation; Test of Buchner's method of purifica- 
tion by hydrogen gas . . 450 

Purification by other gaseous reagents; Mallet's method of making 

chemically pure aluminium 451 

Ive Verrier's experiment with alkaline fluoride; Use of sodium to re- 
move dissolved gases; Annealing; Hardening 452 

Effect of chilling; Rolling; Beating into leaf 453 

Rolling into tubes by the Mannesmann process; Very fine tubes made 

by Ivins, of Philadelphia; Drawing into wire . . . 454 

Stamping and spinning; Grinding, polishing and burnishing . . 455 

Special polishes for aluminium; Engraving; Cleaning and pickling; 

Mat . . ... 456 

Welding by electricity; Soldering; Requirements of a satisfactory 
solder; Difficulties peculiar to aluminium . . . . 457 

Deville's views on soldering aluminium . . . 458 

Mourey's first practical solders . ... ... 459 

Bell Bros. ' method of using Mourey's solders ..... 461 

Frishmuth's solders; Schlosser's solder for dental work .... 462 

Bourbouze's solder of aluminium and tin; Thowless' patent solder . 463 

Patents of Sellon and Sauer; Novel's alloys for soldering . . . 464 
Land's mechanical method; Aluminium specially prepared for soldering 465 
Special fluxes recommended; Roman's solder . . . 466 

The "Alsite" solder; The phosphorized solder of Mr. Joseph Richards, 

of Philadelphia 467 

Success of Richards' phosphorized solder; Methods of using . . . 468 
Coating other metals with aluminium; Veneering as practiced by M. 

Sevrard . . 469 

Dr. Winckler's experience . 470 



Dr. Gehring's method of aluminizing; Broadwell's process for coating 
sheet-iron ...... 471 

Plating other metals on aluminium; Gilding and silvering by Mourey 

and others 472 

Veneering with silver or platinum; Rolling together aluminium and 
copper sheet; Wegner and Guhr's methods for electro-plating and 
coloring; Neeson's methods; The uses of aluminium . . . .473 
The place of aluminium among the useful metals . ... 474 

The first articles made of aluminium; Use in the French army . . 475 
Recommendations of the German military commission; Use for boats . 476 
The yacht " Vendenesse;" The French aluminium torpedo boat . . 477 
Explorer Wellman's boats and sledges; The yacht "Defender;" Use for 

vehicles and flying machines 478 

Use architecturally and for decorations; Suitability for mine cages 479 

Use in surgery; Dental plates ... . . . 480 

Use for scientific instruments; Captain Gordon's sextant . . .481 

Advantages of aluminium for engineering and electrical instruments; 

Chemical and bullion balances . . 482 

Troemner's assay balances; Use for aluminium for minor coinage . . 483 
Advantages and suitability for coinage; Use for chemical apparatus . 484 
Use for culinary utensils; Its well-established advantages 
Use for table ware; Disadvantages for some articles 
Aluminium constructions; Use in the battery 
Substitution for lithographic stone; Strecker's directions 
Use for flash-light powder; Recipes of Villon and Black 
Miscellaneous uses ........ 



Ai,i,OYS OF Aluminium. 

General remarks; Methods of making the alloys . . . . 492 

Division of the alloys into two classes; Influence of Aluminium on the 

color of other metals ... 493 

Contraction in alloying; Influence of aluminium on the specific gravity 

of other metals . 494 

Alloys with copper and iron to be considered separately; Alloys with 

antimony 495 

Unusual property of one antimony alloy; Roche's unstable alloy . . 496 
Aluminium and arsenic; Aluminium and bismuth; Aluminium and 

boron . . 497 

Aluminium and cadmium; Aluminium and calcium .... 498 

Aluminium and chromium; Langley's alloy 499 

Aluminium and cobalt; Lejeal's test; Aluminium and gallium . . 500 

xxxii CONTENTS. 


Aluminium and gold; Roberts-Austin's tests; Niirnberg gold; Andrews' 

experiments .... ....••• 501 

Peculiar properties of the violet alloy discovered by Roberts-Austin . 502 

Aluminium and lead; Cupellatiou of aluminium 503 

Aluminium in anti-friction and type metal; Aluminium and magnesium. 504 

Aluminium and manganese; Cowles' alloy • 505 

Aluminium and mercury; Deville's mistake; Simple experiment by the 

author; Cailletet's method of amalgamation ..... 506 

Joule's electrical experiment; Gmelin's observation; Method of amal- 
gamation by Watts; Bailie and Fury's study of the subject . 507 
Composition and properties of a definite aluminium amalgam . . 508 
Action of air, water, acids and other metals, on aluminium amalgam; 

Aluminium and molybdenum ........ 509 

Aluminium and nickel; Tissier's tests . ..... 510 

I^ejeal's alloy with nickel ; The Pittsburgh Reduction Co.'s alloy; 

Aluminium- nickel-copper alloys 511 

Alloys of Sauvage, Baudrin, Webster; "Neogen" . . . 512 

Webster's patented alloys ... ...... 513 

" Lechesue; " Method of production and properties . . 515 

Cowles Bros.' "Aluminium Silver" and "Hercules Metal;" Andrew's 

alloys; Aluminium and phosphorus ... . . 516 

Aluminium and platinum; Aluminium and silicon ; Alloys formed 

while melting aluminium in crucibles ...... 517 

Influence of silicon on commercial aluminium . . . . 518 

Aluminium and selenium; Aluminium and silver ..... 519 

Silver alloy for dental plates; " Tiers Argent " 520 

Aluminium and sodium; Aluminium and tellurium .... 521 

Aluminium and tin; Bourbouze alloy . 522 

Decomposition of some aluminium-tin alloys 523 

Aluminium deleterious to tin-foil; Heycock and Neville, and Minet, on 

the fusing points of aluminium-tin alloys ...... 524 

Aluminium and titanium; Wohler's alloy ...... 525 

Levy's alloy; Alloy made by the Pittsburgh Reduction Company . . 526 
Le Verrier's tests of titanium alloys; Aluminium and tungsten; Tests 

by he Verrier .... . ,527 

Aluminium and zinc; How zinc crept into some early commercial 

aluminium .... . ... . 528 

Aluminium hardened by zinc and copper; Use of aluminium in the 

galvanizing bath and in Parke's desilverizing process .... 529 

Aluminium-zinc-copper alloys; Aluminium brasses 530 

Baur and Farmer's aluminium brasses; Tests of Cowles Bros', brasses. 531 

Professor Tetmayer's tests of aluminium brasses 532 

Richards' bronze .... 533 


Ai,uminium-Coppe;r Ai<i,oys. 


Division into two classes; General properties 534 

Melting points; Indications of chemical alloys 535 

Alloys containing small amounts of copper . ... 536 

Aluminium hardened by copper; Tests of these alloys by Captain 

Julien ... 537 

Aluminium hardened by German silver; Alloys containing small 
amounts of aluminium; Aluminium bronzes ... . 538 

History of the aluminium bronzes 539 

Constitution of aluminium bronzes; Are they chemical combinations? 540 
Morin's arguments to prove that the bronzes are chemical alloys . . 541 
Action of small amounts of aluminium on bronze, brass and copper . 542 
The reducing action of aluminium on dissolved oxides in molten 

metals . 543 

Bronzes made from alumina versus those made with metallic alu- 
minium 544 

Method of preparing the bronzes industrially . .... 545 

Influence of impurities in the aluminium or copper on the properties 
of the bronzes ...... .... 546 

Re-melting the bronzes improves their quality ..... 547 

Fluidity of the bronzes when melted; Casting the bronzes . . . 548 
Thomas D. West on " Casting Aluminium Bronze " . . 549 

Shrinkage of the bronzes in cooling; Color . . £52 

Specific gravity of the bronzes; Hardness .... . 553 

Transverse and compressive strength ... . 554 

Tensile strength; Lechatelier's tests . . . . . 555 

Tests of tensile strength by Deville, Strange, at the Watertown Arsenal 
and Washington Navy Yard ......... 556 

Details of tests of Cowles Bros', bronzes . 557 

Diagram showing comparison of the strength of wrought-iron, tool- 
steel, and two aluminium bronzes ........ 558 

Professor Tetmayer's tests of the strength of the bronzes . . . 559 
Diagram showing the variation of strength and ductility of the bronzes 
with increasing proportions of aluminium; Le Chatelier's test of the 

strength at various temperatures 560 

Annealing and hardening the bronzes; Forging, hammering and 

rolling -561 

Filing, chipping and planing the bronzes 562 

Anti-friction qualities of the bronzes 563 

Conductivity for heat and electricity; Resistance to corrosion . . 564 
Suitability of the bronzes for marine construction, chemical purposes 
and statuary . 566 

xxxiv CONTENTS. 


Various uses to which the bronzes are applied; Suitability for heavy 

guns . ... 

Suitability for propeller blades 
Brazing and soldering the bronzes 
Silicon-aluminium bronze . 
Phosphor-aluminium bronze; Boron-aluminium bronze 

. 568 
. 569 
. 570 
. 571 


Aluminium-Iron Ahoys. 

Iron in commercial aluminium . . ...... 672 

Alloys made by Tissier Bros., Michel, and Calvert and Johnson . . 573 
Ferro-aluminium ; Manufacture directly from alumina or from the 

metals . . ... . . ■ 574 

Aluminium in commercial steel, wrought-iron and pig-iron; Effect of 

aluminium on the properties of steel ....... 575 

Faraday's experiments with Wootz steel 576 

Mitis castings; Use of aluminium in casting soft steel . . . 577 

Experiments of Gilchrist and R. W. Bavenport .... 579 

Spencer's experiments on casting steels . ..... 580 

' ' Steel-aluminium ' ' for use in steel ; Effect of aluminium on wrought- 
iron . ... 581 

History of the Mitis process 582 

Rise and progress of the Mitis process; Raw material used . . . 583 
Composition of a melt of mitis metal ; Analyses of mitis castings . . 584 
The process of making mitis castings ....... 585 

Conduct of the process at the Chester Steel Casting Works . . . 586 
Explanation of the action of aluminium in making mitis castings; Ost- 

berg's views ..... ..... 587 

R. W. Davenport's explanation inadequate . .... 588 

The author's explanation of the increased fluidity of the melt . . 589 
Why mitis castings are without blow-holes . .... 590 

Discussion of Davenport's experiment and Howe's explanation . . 592 
Professor Le Verrier's explanation ; Professor Arnold's experiment . 593 
Summary of the author's views; Influence of aluminium in the puddling 

furnace . . ... 594 

Influence of aluminium on cast-iron . ...... 595 

Systematic investigation of this subject by Messrs. Keep, Maybery and 

Vorce . . .... .... 596 

General outline of Keep's tests 597 

Effect of aluminium on the solidity of the castings; Does the aluminium 

remain in the iron ? 59g 

Effect of aluminium on the transverse strength of castings . . . 599 
Effect on the elasticity of the castings 600 



Effect on the grain of the castings 601 

General effect of aluminium on the fluidity of cast-iron; Shrinkage . 602 
Effect on the hardness of the castings . ... . 603 

Practical advantages in pouring the castings; General testimony of ex- 
perienced foundrymen . . . .... 604 

Rationale of the action of aluminium on cast-iron ... . 605 


Analysis of Ai,uminium and Aluminium Alloys. 

Ingredients to be determined; Qualitative tests to precede the quantita- 
tive analysis ... 606 

Qualitative blow-pipe tests; The specific gravity as an indication of 
purity . ... .... 607 

Schulze's test of the purity of commercial aluminium; Determination 
of silicon . . . . .... 608 

Separation of combined and graphitic silicon 609 

Regelsberger's method for dissolving and estimating silicon . . 610 

Determination of iron and aluminium .611 

Separations of iron from aluminium 613 

Determination of lead; Determination of copper; Determination of zinc 614 
Determination of tin; Determination of silver; Determination of sodium 615 
Determination of chlorine; Detemiination of carbon; Determination of 

fluorine . . 616 

Determination of titanium; Determination of chromium . . . 617 
Analysis of ferro-aluminiums; Difficulty of estimating aluminium prop- 
erly; Yates' method of procedure 618 

Chancel's separation of iron from aluminium; Peter's modification of 

Chancel's separation; Thompson's experience 619 

Satisfactory separation of iron from aluminium by potassium cyanide; 

Precipitation of iron by tri-methylamine 620 

Blair's procedure for separating iron from aluminium . . . 621 

Electrolytic separations of iron from aluminium; Classen's method; 
Professor Smith's method ... . . . 622 

Stead's method of separating iron from aluminium; Analysis of alu- 
minium bronzes ... ..... 623 

Determination of tin, lead and copper in the bronzes .... 624 

Determination of iron, aluminium, zinc, cobalt and nickel . . . 625 
Analysis of aluminium-zinc alloy; Analysis of aluminium-tin, alumin- 
ium-silver, aluminium-nickel, aluminium-manganese and aluminium- 
lead alloys ... 626 

General remarks on the usefulness of making qualitative tests prepara- 
tory to a quantitative analysis . 627 

Index ... ... 629 




Pliny* tells us that alumen was the name of several kinds of 
salts used in dyeing, a white kind for dyeing bright colors and 
a dark kind for dull colors ; that all alumen had an astringent 
taste and was colored black by the juice of the pomegranate 
(a reaction caused by iron). From all we can learn in the 
writings of Pliny and Columella, it appears that the white 
alumen must have been a mixture of aluminium sulphate with 
more or less sulphate of iron. Why it was called alumen is 
not certainly known, but it was probably derived from lumen, 
light, in allusion to its brightening the colors when used in 

About the eighth century a very pure alumen was made in 
Rocca, near Smyrna, and hence called alumen Rocca, or in 
modern times, rock alum. This was, indeed, a pure sulphate 
of aluminium with sulphate of potash; in fact, our modern 
alum, diflfering from the ancient alumen in that it contained 
potassium, and was manufactured with care instead of being 
merely a natural substance. The old name was, however, still 
attached to the new product, and included in its scope the 
vitriols, that is, the sulphates of iron and copper. Even in the 
thirteenth century alum and vitriol vvere named together as re- 
lated substances. 

* Book 35, chapter 15. 



Paracelsus* in the sixteenth century first separated alum and 
vitriols, on the ground that the base of the vitriols was metallic 
and that of alum earthy. 

Ettmiillerf discovered in 1684 that "alum is obtained by 
acting on clay with sulphuric acid." 

Stahl in 1702 expressed clearly his conviction that the un- 
known base of alum was of the nature of lime or chalk, and for 
fifty years thereafter this was the generally accepted idea. 

Hoffman $ twenty years later announced that the base of 
alum appeared to him to be a true, distinct earth. GeofiTroy, 
in 1728, Hellot, in 1739, and Pott, in his famous Lithogeognosia, 
in 1746, all reiterated this statement with increasing certainty, 
until it was finally generally received as an accepted fact, and 
the last named chemist gave it the name of thonichte erde, or 
terre argilleuse. 

In 1754 Marggraff,§ in three able dissertations on alum and 
its earth, showed that this earth is certainly a distinct substance, 
that it exists in all natural clays and can be extracted therefrom 
by sulphuric acid, that the part of the clay not touched by the 
acid is silica, and that, therefore, the purest white clay contains 
only silica and the earthy base of alum. These memoirs settled 
conclusively the composition of clay, and give rise to the term 
argil or argil pier as the name of the earthy base of alum. It 
was thus called until 1761, when Morveau,|| intent on revising 
chemical nomenclature, decided that since alum was called sel 
alumineux, the proper name for its base should be altimine, 
thus avoiding all confusion with argille and terre argilleux, the 
terms designating clay and clayey earth. In German, however, 
the earth of alum is still called thonerde — clay earth ; in English, 
the French alumine was adhered to until about 1820, when it 
was anglicised into alumina. 

* Book 3, p. 64. 

t Chymia Rationalis ac Experimentalis. 

X Observationem physico-chemicarum Selectiorium, 1722, 

§ Opusc. Chim. de Marggraff, I, p. 8. 

II Dictionnaire de Physique, Article "Alun." 


Returning to the time of Marggraff, we find soon after the 
pubHcation of his paper indications that chemists suspected 
this newly-identified earth to be a compound body, and the 
search after aluminium had truly begun. 

Macquer, * in 1758, wrote, "the earth of alum is white, infus- 
ible, and I suspect that it has a relation more or less distant 
with the metallic earths." 

Baron, f a professor of chemistry in Paris, is our first re- 
corded experimenter on the isolation of aluminium. In 1760 
he communicated a memoir to the Academy in which he said : 
" Marggrafif of Berlin has shown us what the earth of alum is 
not, without showing very definitely what it is. I believe the 
base of alum to be of a metallic nature, for the following 
reasons: i. It has almost no properties in common with the 
known earths. 2. It has analogies with the metallic earths, e. g., 
its astringent salts." Baron then proceeds to say that he had 
tried all known methods of reducing this base, but without re- 
sult, and concludes by saying, " If I had been fortunate enough 
to reduce the base of alum to a metal, no other argument would 
be needed, and that which to-day I conjecture would be a de- 
monstrated fact. I am far from regarding the problem as an 
impossibility. I think it not too venturesome to predict that 
a day will come when the metallic nature of the base of alum 
will be incontestably proven." 

Baron does not record the manner in which he tried to re- 
duce alumina, but it is probable that he mixed it with carbon 
or some organic substance, with salt or soda for flux, and 
heated as highly as possible in a charcoal fire. We neverthe- 
less have his word that he tried all the methods of reduction 
then known. 

Lavoisier % regarded it as highly probable that alumina was 
the oxide of a metal, the aflSnity of the metal for oxygen being 

* Memoires de Paris, 1758. 

t Memoires de 1' Acad. Royale, April l6, 1760, p. 274. 

t Journal de Physique, May, 1782. 


SO strong that neither carbon nor any of the reducing agents 
then known was able to overcome it. 

Ruprecht and Tondi,* two Austrian chemists, repeated 
Baron's experiments in 1790. They thought that hitherto the 
reducing agent had not been intimately enough mixed with the 
alumina, and that the temperature used had not been sufficiently 
high, so they mixed alumina very intimately with charcoal dust, 
made it into a paste with oil, and spread it on the inner walls 
of a Hessian crucible, which was finally filled up with powdered 
charcoal and a layer of bone ash placed on top. Putting the 
crucible in a forge, a strong fire was kept up for three hours. 
As a result, small metallic particles were found on the inner 
sides of the crucible. These they supposed to be the metallic 
base of alumina. 

Sav4resi,f in Italy, and Klaproth and Karsten, in Germany, 
duplicated this experiment with similar results, and then 
analyzed the globules of metal obtained. They found them to 
be phosphide of iron, the iron coming from the charcoal used 
and the phosphorus from the bone ash. Professor Klaproth 
referred to this incident as " the pretended metallization of the 
earths," and said further, " if there exists an earth which has 
been put in conditions where its metallic nature should be dis- 
closed, if it had such, an earth exposed to experiments suitable 
for reducing it, tested in the hottest fires by all sorts of 
methods, on a large as well as on a small scale, that earth is 
certainly alumina, yet no one has yet perceived its metalli- 

Lavoisier succeeded in melting alumina in a charcoal fire fed 
by pure oxygen. Dr. Hare, in 1802, melted it with the oxy- 
hydrogen blow- pipe to a white enamel, but noticed no signs of 
reduction to metal. 

In 1807 Sir Humphry Davy attacked the alkaline earths 
with his great battery which had isolated potassium and so- 

* Allgemeiner Zeitung, Nov., 1 790. 
t Annales de Chimie, 1791, »., 254. 


dium. His first experiment, mixing alumina with red oxide of 
mercury, and passing the electric current through this in con- 
tact with metallic mercury, gave no result. He then melted 
alumina with potash, and passed the current from 500 plates 
through it. The receiving blade received a metallic coating, 
and on dipping it into water the latter was decomposed, and 
afterwards the solution was proved to contain alumina. An 
experiment with soda gave a similar result. Davy had in fact 
electrolyzed the aluminates of potassium and sodium, and ob- 
tained metallic alkali and aluminium, but he was unable to 
isolate the latter by any means he could devise. 

Davy next tried to reduce alumina by heating it with metal- 
lic potassium. Some potassium oxide was formed, showing 
some reduction had taken place, but he could find no metal. 

In 1809 Davy had a new battery of 1000 plates, and with it 
fused iron to whiteness in the arc in contact with alumina. 
The iron became somewhat whiter, and when dissolved in acids 
showed that it contained aluminium. The fact was thus estab- 
lished that alumina can be decomposed while fiuid in the electric 
arc, and its metal alloyed with iron. 

Davy next mixed alumina with potassium and iron filings, 
hoping that the iron would collect any metal reduced from the 
alumina. On melting this mixture a button resulted which 
was white and harder than iron, and was undoubtedly an alloy 
of iron and aluminium, but Davy could not separate the two 
metals. In concluding the recital of his experiments he said, 
" Had I been fortunate enough to isolate the metal after which 
I sought, I would have given it the name alumium. 

In making this suggestion it is perfectly plain that Davy in- 
tended this word to represent the metal from alum, simply 
starting with alum, and adding ium as the proper termination. 
Objections were very soon thereafter made to this proposed 
name, not to the termination ium, which was considered abso- 
lutely proper, but to the root or stem of the word. It was 
maintained by French, German and Swedish writers that the 
name of the new metal should be derived from its oxide, and 


that the stem of the word should therefore be alumin, and 
thence the name aluminium. Davy was influenced by these 
criticisms to the extent of changing in 1812 to alumin-um, but 
no writers, except a very few English and, in recent years, some 
Americans, have used this spelling. 

Prof. Benj. Silliman* was in 1813 repeating Hare's experi- 
ment of fusing alumina with the oxy-hydrogen blowpipe. The 
alumina was supported on charcoal, and he noticed small me- 
tallic globules rolling and darting out from under the fused 
mass and burning with a bright light. These globules could 
have been nothing else than aluminium reduced from fluid 
alumitia by charcoal. The metal burnt, however, almost in- 
stantly, and no globules could be obtained. 

The Swedish chemist Oersted believed, in 1824, that he had 
isolated aluminium. He used anhydrous aluminium chloride 
as a starting point, and heated it with potassium amalgam, 
thinking that the potassium would reduce the chloride and 
leave an amalgam of aluminium and mercury, from which the 
latter could be distilled away. Potassium, however, when 
amalgamated with mercury is not powerful enough to reduce 
aluminium chloride, although it can reduce it if used alone, and 
so Oersted missed, by a very little, the honor of first isolating 

Berzelius was also a worker on the aluminium problem. He 
wrote to Sir Humphry Davy, in 1809, that he had reduced 
alumina by heating it with carbon in the presence of iron, but 
it is probable that he found himself mistaken, as he does not 
refer to this matter in any of his later writings ; once, however, 
he came within a very little of succeeding. He had studied 
the composition and formula of the mineral cryolite, had made 
it artificially, and also the corresponding potassium salt. It 
then occurred to him that this latter salt might be reduced by 
metaUic potassium. He made the experiment in a crucible, 
carefully washed the fusion with water, but found no metal. It 

* Memoirs Connecticut Acad, of Arts and Science. Vol. I, 1813. 


is quite evident that his sole mistake was in using an excess of 
potassium, which, after the reduction, gave on sokition in water 
a caustic alkah sokition, which immediately dissolved all the 
reduced aluminium. Had he used an excess of the fluoride 
salt, the story would almost certainly have read difTerently, be- 
cause this reduction has since then been repeatedly performed. 
The date of this experiment was 1825. 

In 1827, Frederick Wohler, Professor of Chemistry at the 
University of Gottingen, after repeating Oersted's experiment 
with unsatisfactory results, modified the experiment by using 


pure potassium as the reducing agent, instead of its amalgam. 
He obtained a grey, metallic powder, which was finely-divided 
aluminium. He was unable to melt this powder into a button, 
so as to determine its properties en masse, but he described many 
of its chemical properties, and produced from it some alumin- 
ium compounds which had not been before made, such as the 
sulphide and arsenide. This powder was not absolutely pure 
aluminium, as it contained a little platinum from the crucible 
in which the reduction was performed, and some potassium, 
which lessened its resistance to chemical agents. At that time 


no further use was made of these facts. Later, in 1845, on 
making vapor of aluminium chloride pass over potassium 
placed in platinum boats, Wohler obtained the metal in small 
malleable globules of metallic appearance, from which he was 
able to determine the principal properties of aluminium. But 
the metal thus obtained was scarcely as fusible as cast iron, 
without doubt because of the platinum with which it had 
alloyed during its preparation. In addition to this, it decom- 
posed water at io6°C, from which we suppose that it was still 
impregnated with potassium or aluminium chloride. It is to 


H. St. Claire Deville that the honor belongs of having, in 1854, 
isolated aluminium in a state of almost perfect purity, deter- 
mining its true properties. 

Thus, while aluminium had been isolated in 1827, for 
eighteen years its properties en masse were unknown, and it 
was only at the end of twenty-seven years after its discovery 
that the true properties of the pure metal were established by 
Deville. The second birth of aluminium, the time at which it 
stepped from the rank of a curiosity into the number of the 
useful metals, dates from the labors of Deville in 1854. If 


Wohler was the discoverer of aluminium, Deville was the 
founder of the aluminium industry. 

In commencing researches on aluminium, Deville, while he 
applied the method of Wohler, was ignorant of the latter's results 
of 1845. Besides, he was not seeking to produce aluminium 
that he might turn its valuable properties to practical account, 
but that it might serve for the production of aluminium prot- 
oxide (AlO), which he believed could exist as well as ferrous 
oxide (FeO). The aluminium he wished to prepare would, he 
thought, by its further reaction on aluminium chloride, form 
aluminium proto-chloride (AlCP), from which he might derive 
the protoxide and other proto-salts. But on passing vapor of 
aluminium chloride over the metallic powder formed by reduc- 
tion by potassium, this proto-chloride was not thus produced; 
he obtained, inclosed in a mass of aluminium-potassium chlo- 
ride (AICI3.KCI), fine globules of a brilliant substance, ductile, 
malleable, and very light, capable of being melted in a mufifle 
without oxidizing, attacked by nitric acid with diflSculty, but 
dissolved easily by hydrochloric acid or caustic potash with 
evolution of hydrogen. 

Deville troubled himself no more about the proto-salts of 
aluminium, but, recognizing the importance of his discovery, 
turned his attention to preparing the metal. He was at this 
time Professor of Chemistry in the Ec^le Normale, Paris ; his 
salary was but 3000 francs, his estate was small, and he was 
practically without the means of doing anything further. 

On Monday, February 6, 1854, Deville read at the .seance of 
the Academy a short paper entitled "Aluminium and its Chemi- 
cal Combinations," in which he explained the results of this 
experiment as showing the true properties of aluminium, 
and also furnishing a method of purifying it, and declared his 
intention of commencing immediate search for a process which 
could be economically applied on a commercial scale. M. 
Thenard, at the close of the communication, remarked that such 
experiments ought to be actively pursued, and that, since they 
were costly, he believed the Academy would hasten the accom- 


plishment of the work by placing at Deville's disposal the neces- 
sary funds. As the outcome of this, the Academy appointed 
Deville one of a committee to experiment on producing alumin- 
ium, and 2000 francs were placed at his disposal for the work. 

It was on the occasion of the reading of this paper that M. 
Chenot addressed a note to the Academy on the preparation 
of aluminium and other earthy and alkaline metals, in which he 
claimed, in some regards, priority for his inventions. (See 
further under "Reduction by Carbon.") This note was re- 
served to be examined by a commission appointed to take 
notice of all communications relative to the production of 

With the funds thus placed at Deville's disposal, he experi- 
mented at the Ecole Normale for several months. As potas- 
sium is very dangerous to handle, cost then 900 francs a kilo, 
and gives comparatively but a small return of aluminium, De- 
ville, in view of the successful work of Bunsen on the electric 
decomposition of magnesium chloride, tried first the reduction 
of aluminium chloride by the battery. On March 20, 1854, 
Deville announced to the Academy in a letter to Dumas that 
he had produced aluminium without alkaline help, and sent a 
leaf of the metal thus obtained. At that time Thenard, Boussin- 
gault, Pelouze, Peligot, and later, de-la-Rive, Regnault, and 
other well-known scientists, shared the honor of assisting in the 
laboratory experiments. Deville sent, in the following May, a 
mass of five or six grammes weight to Liebig, making no secret 
of the fact that it was reduced by the battery ; while Balard at 
the Sorbonne, and Fremy at the Ecole Polytechnique, publicly 
repeated his experiments and explained them in all their de- 
tails. Although these experiments succeeded quite well, yet 
because of the large consumption of zinc in the battery used 
the process could evidently not be applied industrially, and 
Deville felt obliged to return to the use of the alkaline metals. 

Towards the middle of 1854 Deville turned to sodium, with- 
out a knowledge of those properties which render it so prefer- 
able to potassium, but solely because of its smaller equivalent 


(23 to that of potassium 39) and the greater cheapness of soda 
salts. He studied the manufacture of sodium, with the aid of 
M. Debray, in his laboratory at the Ecole Normale, and their 
experiments were repeated at Rousseau Bros.' chemical works 
at Glaciere, when they were so successful that Rousseau Bros, 
very soon put metallic sodium on the market at a much reduced 
price. It is said that while metallic sodium was a chemical 
curiosity in 1855, costing something like 2000 francs a kilo, its 
cost in 1859 is put down at 10 francs. Deville carried this pro- 
cess to such perfection that for twenty-five years it remained 
almost precisely at the status in which he left it in 1859. In 
order to still further cheapen aluminium, Deville busied himself 
with the economic production of alumina, which gave later a 
lively impulse to the cryolite and bauxite industries. 

On August 14, 1854, Deville read a paper before the Acad- 
emy describing his electrolytic methods at length (see under 
"Reduction by Electricity"), showing several small bars of the 
metal, and also stating some of the results already achieved by 
the use of sodium, but not going into details, since he believed 
that numerous analyses were necessary to confirm these results 
— which he was unable to have made with the funds at his dis- 
posal. He also stated that the desire to show, in connection 
with his assertions, interesting masses of the metal, alone pre- 
vented the earlier publication of the methods used. Several 
days before this, Bunsen published in Poggendorif's Annalen 
a process for obtaining aluminium by the battery, which re- 
sembled Deville's method, but of which the latter was ignorant 
when he read his paper. Thus it is evident that the isolation 
of aluminium by electrolysis was the simultaneous invention of 
Deville and Bunsen. 

After reading this paper, Deville caused a medal of alumin- 
ium to be struck, which he presented to the Emperor Na- 
poleon III. The latter, looking forward to applying such a 
light metal to the armor and hemlets of the French Cuirassiers, 
immediately authorized experiments to be continued at his own 
expense on a large scale. This anticipation ultimately proved 


impracticable, but the ambition in which it was bred was caused 
for once to minister to the lasting benefit of mankind. Deville, 
however, about this time accepted, in addition to his duties as 
professor at the Ecole Normale, a lectureship at the Sorbonne 
(where he afterwards obtained a full professorship), and it was 
not until March of the next year that the experiments at the 
cost of the Emperor were begun. 

It was about August, 1854, that two young chemists, Chas. 
and Alex. Tissier, at the suggestion of Deville, persuaded M. 
de Sussex, director of a chemical works at Javel, to let them 
experiment in his laboratory (of which they had charge) on 
the production of sodium. 

Towards the commencement of 1855, Deville took up the in- 
dustrial question, the Emperor putting at his disposition all the 
funds necessary for the enterprise, and in March the investigator 
went to work and installed himself at the chemical works at 
Javel in a large shed which the director, M. de Sussex, kindly 
put at his service. 

The investigations were carried on here for nearly four 
months, ending June 29th, and the process elaborated was an 
application on a large scale of the experiments he had made at 
the expense of the Academy, which he described in his paper 
of August 14, 1854, and by which he had been able to obtain 
a few pencils of metal. In this work such success attended his 
efforts that on June 18 Deville presented to the Academy 
through M. Dumas large bars of pure aluminium, sodium, and 
masses of aluminium chloride. The members and large audi- 
ence were loud in their admiration and surprise at the beauty 
of the metal. Dumas stated that the experiments at Javel had 
put beyond a doubt the possibility of extracting aluminium on 
a large scale by practical processes. Deville's paper was then 
read, describing all his processes in detail, and concluding with 
the following words : " After four months of work on a large 
scale, undertaken without responsibility on my part, and, in 
consequence with the tranquility and repose of mind which are 
so often wanting to the investigator, without the preoccupation 


of expense, borne by His Majesty the Emperor, whose gen- 
erosity had left me entire liberty of action, encouraged each 
day by distinguished men of science, I hope to have placed the 
aluminium industry on a firm basis." 

It was the metal made at this time at Javel which was ex- 
hibited at the Paris Exposition in 1855. In the Palais de I'ln- 
dustrie, among the display from the porcelain works at Sevres, 
were ingots and some manufactured objects. The first article 
made of aluminium was, in compliment to the Emperor, a baby- 
rattle for the infant Prince Imperial, for which purpose it must 
have served well, because of the sonorousness of the metal. 

After terminating these experiments Deville continued work- 
ing at the Ec61e Normale, the Emperor defraying his expenses, 
until April, 1856. The memoir published in the Ann. de Chim. 
et de Phys., April, 1856, contains, besides the results obtained 
at Javel, the improvements devised in the meantime. 

It appears that when Deville first went to Javel, he had for 
assistants the Tissier Brothers, who were charged by M. de 
Sussex to give him all the aid they could. Since the previous 
autumn the Tissiers had been experimenting on sodium fur- 
naces, and now, in concert with Deville, they drew up plans for 
furnaces, and aided in devising other apparatus. Under these 
circumstances the furnace for the continuous manufacture of 
sodium in cylinders was devised, which the Tissiers claim De- 
ville strongly advised them to make their property by patenting, 
asking only from them the use of it for his experiments. So, 
immediately after the experiments were ended, in July, the 
Tissiers patented the furnace in question, and, leaving Paris, 
took charge of M. Chanu's works at Rouen. On the other 
hand, Deville always reproached them for acting in bad faith. 
He says that after having assisted for about two months in set- 
ting up his apparatus, being forced to leave the works because 
of misunderstandings between them and M. de Sussex, they 
were admitted to his laboratory at the Ecole Normale, and 
initiated by him into the knowledge of all those processes 
which they made use of afterwards, then suddenly left, taking 


drawings of furnaces, details of processes, etc., which they not 
only made free use of, but even patented. However, whichever 
party was in the right (and those who comprehend the char- 
acter of Deville can hardly doubt which was), the fact stands 
that in July, 1855, M. Chanu, an honorable manufacturer of 
Rouen, founded a works in which Deville's processes were to 
be applied, and intrusted the direction of it to the Tissier 

The history of the works at Rouen is thus described by the 
Tissiers in their book on aluminium, of which we shall speak a 
little further on : 

"la July, 185s, Messrs. Maletra, Chanu, and Davey, of 
Rouen, formed a company to produce aluminium, and we were 
intrusted with the organization and special charge of the in- 
dustry. The commencement was beset with difficulties, not 
only in producing, but in using the metal. It then sold at $200 
per kilo, the price being an insurmountable obstacle to its em- 
ployment in the arts. The small capital at our disposal was 
not enough to start the industry, to pay general expenses, and 
the losses occasioned by the many experiments necessary. On 
February 28, 1856, the society was dissolved. In April of the 
same year, M. William Martin, struck by the results already 
obtained and sanguine of greater success, united with us. From 
that time daily improvements confirmed M. Martin's hopes, 
and in 1857 the works at Amfreville-la-mi-Voie, near Rouen, 
sold the metal at $60 per kilo ($2 per oz.). The laboratory 
of this works was devoted to researches on everything concern- 
ing the production and application of aluminium. M. Martin 
has our sincere gratitude for the kindness with which he so 
willingly encouraged and contributed to the progress of the 
manufacture of this wonderful metal." 

The process ultimately used at Amfreville was the reduction 
of cryolite by sodium, but the enterprise was not a permanent 
success, and after running for a few years it was abandoned and 
the works closed. 

Returning to Deville, we find that after leaving Javel one of 


the first subjects he investigated was the use of cryolite for 
producing aluminium. The researches made with the aid of 
MM. Morin and Debray were published in the memoir of April, 
1856, and became the basis of the process carried out by the 
Tissiers at Rouen. Besides this, Deville perfected many of the 
details of a practical aluminium plant, with the result that in 
the spring of 1856 he united with Messrs. Debray, Morin, and 
Rousseau Bros, (the latter manufacturers of chemicals at Gla- 
ciere, in whose works aluminium had been made since the 
middle of 1855) and put up new apparatus in the works at 
Glaciere, the company furthering the work entirely at their own 
cost. This enterprise lasted for more than a year, during which 
a number of processes were tried and continued improvements 
made, so that towards August of the same year aluminium 
was put on the market in Paris at 300 francs a kilo, being one- 
third what it cost a year previous. 

Finally, in April, 1857, the little works at Glaciere, a suburb 
of Paris, in the midst of gardens and houses, and turning into 
the air fumes charged with chlorine and salts, was obliged by 
reason of general complaints to stop making aluminium. The 
plant was moved to Nanterre, where it remained for some years, 
under the direction of M. Paul Morin, being on a scale four 
times as large as the actual demand. Afterwards part of the 
plant was moved to the works of H. Merle & Co., at Salindres, 
and later on the whole plant, where the manufacture was until 
recently carried on by Pechiney & Co. The works at Nan- 
terre were really the only "aluminium works" built by Deville, 
the others were plants installed at general chemical works ; but 
these at Nanterre were built by the united efforts of Deville, 
his brothers and parents, and a few personal friends. Among 
those who aided Deville, especially in the problems which the 
new industry presented, he speaks warmly of Messrs. d'Eichtal, 
Lechatelier, and Jacquemont. 

In 1858 the Tissiers wrote and published- a small work 
entitled " Recherches sur 1' Aluminium." which, in view of what 
Deville could have written about the subject, was a decided 


misrepresentation of the results which had been thus far accom- 
plished. Deville thought that the industry was yet too young 
to merit any sort of publication, yet he naively writes in his 
work "De 1' Aluminium," in 1859, "I will sincerely acknowl- 
edge that my writing is a little due to my pride, for I decided 
to take the pen to speak of my work, only to avoid seeing it 
belittled and disfigured as it has been lately in the book written 
by MM. Tissier." 

Deville published his book in September, 1859, and he con- 
cludes it with these words : " I have tried to show that alumin- 
ium may become a useful metal by studying with care its 
physical and chemical properties, and showing the actual state 
of its manufacture. As to the place which it may occupy in 
our daily life, that will depend on the public's estimation of it 
and its commercial price. The introduction of a new metal 
into the usages of man's life is an operation of extreme diffi- 
culty. At first aluminium was spoken of too highly in some 
publications, which made it out to be a precious metal, but 
later these estimates have depreciated even to the point of con- 
sidering it attackable by pure water. The cause of this is the 
desire which many have to see taken out of common field-mud 
a metal superior to silver itself ; the opposite opinion established 
itself because of very impure specimens of the metal which were 
put in circulation. It seems now that the intermediate opinion, 
that which I have always held and which I express in the first 
lines of my book, is becoming more public, and will stop the 
illusions and exaggerated beliefs which can only be prejudicial 
to the adoption of aluminium as a useful metal. Moreover, 
the industry, established as it now is, can be the cause of loss 
to no one ; as for myself, I take no account of the large part of 
my estate which I have devoted, but am only too happy if my 
efforts are crowned with definite success, in having made fruit- 
ful the works of a man whom I am pleased to call my friend, 
the illustrious Wohler." 

Contemporary with the early labors of Deville, among the 
numerous chemists and metallurgists investigating this attract- 


ive field, we find Dr. Percy in England and H. Rose in Ger- 
many, whose experiments on the reduction of cryolite by 
sodium were quite successful, and are herein described later on. 

As early as 1856 we find an article in an American magazine * 
showing that there were already chemists in the United States 
spending time and money on this subject. The following is the 
substance of the article alluded to: "Within the last two years 
Deville has extracted 50 to 60 lbs. of aluminium. At the pres- 
ent time, M. Rousseau, the successor of Deville in this manu- 
facture, produces aluminium which he sells at $100 per pound. 
No one in the United States has undertaken to make the metal 
until recently Mons. Alfred Monnier, of Camden, N. J., has, 
according to the statement of Prof. James C. Booth in the 
" Penn. Inquirer," been suceessful in making sodium by a con- 
tinuous process, so as to procure it in large bars, and has made 
aluminium in considerable quantity, specimens of which he has 
exhibited to the Franklin Institute. Mons. Monnier is desirous 
of forming a company for the manufacture of aluminium, and is 
confident that by operating in a large way he can produce it at 
a much less cost than has heretofore been realized. We would 
suggest the propriety of giving aid to this manufacture at the 
expense of the government, for the introduction of a new metal 
into the arts is a matter of national importance, and no one can 
yet realize the various and innumerable uses to which this new 
metal may be applied. It would be quite proper and constitu- 
tional for Congress to appropriate a sum of money, to be ex- 
pended under the direction of the Secretary of the Treasury in 
the improvement of this branch of metallurgy, and in testing 
the value of the metal for coinage and other public use." 

In the next volume of the "Mining Magazine"! there is a 
long article by Mr. W. J. Taylor, containing nothing new in 
regard to the metallurgy of aluminium, but chiefly concerned 
in calculating theoretically the cost of the metal from the raw 

* Mining Magazine,. 1856, vii., 317. 

t Mining Magazine, viii., 167 and 228. Proc. Ac. Nat. Sci., Jan., 1857. 


materials and labor required by Deville's processes, and con- 
cluding that it was quite possible to make it for $ i .00 per pound. 

In 1859 the first aluminium works in England were started at 
Battersea, near London, by C. H. Gerhart, using Deville's pro- 
cess. No details are obtainable respecting the size of this 
works. M. Henri Brivet informs me that they had no connec- 
tion with Bell Bros.' enterprise, started the next year after, and 
that they closed operations about 1865. 

In i860, Bell Bros, established a plant at Washington, near 
Newcastle-on-Tyne, with the immediate co-operation of Deville.' 
In 1862 this company was selling aluminium at 40 shiUings per 
troy pound, and they continued operations until 1874, when 
they closed because the business was no longer profitable.* 

It was probably shortly after 1 874 that the large firm of J. F. 
Wirtz & Co., Berlin, made an attempt to start an aluminium 
works. The project drooped before it was well started ; and it 
is only within the last few years that Germany has possessed a 
flourishing aluminium industry. 

The further we get away from an age the better able are we 
to write the true history of that age. And so as years pass 
since the labors of Wohler, Deville, and Tissier, we are now 
able to see better the whole connected history of the develop- 
ment of this art. Dr. Clemens Winckler gives us a compre- 
hensive retrospect of the field seen from the standpoint of 1879, 
from which we condense the following : -j- " The history of the 
art of working in aluminium is a very short one, so short that 
the present generation with which it is contemporary is in dan- 
ger of overlooking it altogether. The three international exhi- 
bitions which have been held in Paris since aluminium first 
began to be made on a commercial scale form so many memo- 
rials of its career, giving as they did at almost equal intervals 
evidence of the progress made in its application. In 1855 we 
meet for the first time in the Palais de 1' Industrie with a large 

* Personal communication to the writer from Sir Lowthian Bell, 
t Industrie Blatter, 1879; Sci. Am. Suppl., Sept. 6, 1879. 


bar of the wonderful metal, docketed with the extravagant 
name of the ' silver from clay.' In 1 867 we meet with it again 
worked up in various forms, and get a view of the many diffi- 
culties which had to be overcome in producing it on a large 
scale, purifying and moulding it. We find it present as sheets, 
wire, foil, or worked-up goods, polished, engraved, and soldered, 
and see for the first time its most important alloy, aluminium 
bronze. After a lapse of almost another dozen years we see at 
the Paris Exhibition of 1878 the maturity of the industry. We 
have passed out of the epoch in which the metal was worked 
up in single specimens, showing only the future capabilities of 
the metal, and we see it accepted as a current manufacture 
having a regular supply and demand, and being in some re- 
gards commercially complete. The despair which has been 
indulged in as to the future of the metal is thus seen to have 
been premature. The manufacture of aluminium and goods 
made of it has certainly not taken the extension at first hoped 
for in its behalf ; the lowest limit of the cost of manufacture was 
soon reached, and aluminium remains as a metal won by ex- 
pensive operations from the cheapest of raw materials. 

" There are several reasons why the metal is shown so little 
favor by mathematical instrument makers and others. First of 
all, there is the price ; then the methods of working it are not 
everywhere known ; and further, no one knows how to cast it. 
Molten aluminium attacks the common earthen crucible, re- 
duces silicon from it, and becomes gray and brittle. This 
inconvenience is overcome by using lime crucibles, or by lining 
an earthen crucible with carbon or strongly burnt cryoHte clay. 
If any one would take up the casting of aluminium and bring it 
into vogue as a current industrial operation, there is no doubt 
that the metal would be more freely used in the finer branches 
of practical mechanics." 

At the time of Dr. Winckler's writing, the extraction of 
aluminium in France was carried on by Merle & Co., at Salin- 
dres, while the Societe Anonyme de 1' Aluminium, at Nanterre, 
worked up the metal. Both firms were represented at the Ex- 


position of 1878. The prices quoted then were 130 francs a 
kilo for aluminium, and 18 francs for 10 per cent, aluminium 
bronze. From 1874, when Bell Bros.' works at Newcastle-on- 
Tyne stopped operations, until 1882, when a new enterprise 
was started in England by Mr. Webster, the French company 
were the only producers of aluminium. 

Regarding the prospects of the aluminium industry at this 
period, we can very appropriately quote some remarks of the 
late Mr. Walter Weldon, F. R. S., who was a personal friend of 
M. Pechiney (director of the works at Salindres), had given 
great attention to aluminium, and was considered as a first 
authority on the subject. Speaking in March, 1883, before the 
London Section of the Society of Chemical Industry, he stated 
that the only method then practiced for the manufacture of 
aluminium was DeviUe's classical one; that at Salindres, M. 
Pechiney had improved and cheapened it, but that was all the 
progress made in the industry in twenty-five years. Continu- 
ing, Mr. Weldon outlined the possible lines on which improve- 
ments might be made, as : 

1st. Cheapening the production of aluminium chloride, or of 
aluminium-sodium chloride. 

2d. Substituting for these chlorides some other cheaper 
anhydrous compounds of aluminium not containing 

3d. Cheapening sodium. 

4th. Replacing sodium by a cheaper reducing agent. 

Mr. Weldon exhibited the relative cost of the materials used 
in making aluminium, as then carried on by M. Pechiney, as : 

Producing the alumina 10 per cent. 

" " double chloride 33 " 

" " sodium and reducing therewith 57 " 

"Discussing these figures, it is seen that the cost of the alu- 
mina forms but a small item in the cost of the metal, since a 
saving of 50 per cent, in its cost would only cheapen the metal 
5 per cent. A large margin is, however, left in the conversion 


of the alumina into the chloride, and it is here that a large sav- 
ing may be expected, either in cheaper methods of producing 
the chloride or by the substitution of some other cheaper salt 
for the chloride. The only other suitable compounds which 
might replace the latter are the fluoride, iodide, bromide, sul- 
phide, phosphide, or cyanide. The fluoride has been used to 
some extent in the form of cryolite, but, from the impurities in 
the mineral and its corrosive action on the apparatus used for 
reduction, the metal produced is very much contaminated with 
iron and silicon. The bromide and iodide, no matter how pro- 
duced, would always be too costly to replace the chloride. The 
production of the sulphide in a suitable form from which the 
metal can be extracted has thus far not proved a success, and 
even if ever it be thus produced in a suitable condition, it is not 
at all likely to be as cheap a material to use as the chloride. 
The phosphide and cyanide can thus far only be produced from 
the metal itself, and are, therefore, totally out of the question. 
To find a substitute for sodium as a reducing agent has been a 
favorite object of research among chemists for the past thirty 
years, and although every element occurring in any abundance, 
or obtainable at a cheaper rate than sodium, has been tried 
under almost all conditions, yet absolutely nothing has been 
accomplished in this direction that would entitle any one to the 
belief that aluminium can ever be produced chemically without 
the use of sodium. So absorbing to those interested in the 
search for a substitute for sodium has the occupation proved, 
that the effort to cheapen sodium did not receive anything like 
its fair share of attention. Since of the 57 per cent, ascribed 
to the cost of sodium and reduction, 50 per cent, represents the 
sodium, which thus costs about six shillings a pound, there is 
seen to be a very large margin for improvements, since the raw 
materials for a pound of sodium do not cost over i or at most 
2 shillings." 

In 1882 the cost of aluminium was materially cheapened by 
the applications of the inventions of Mr. Webster, which, in ac- 
cordance with the analysis of the problem made by Mr. Weldon, 


consisted principally in the cheap production of alumina and its 
conversion into chloride. Mr. Webster had experimented on 
this subject many years, and in 1881 and 1882 took out patents 
for his processes, and organized the "Aluminium Crown Metal 
Company," located at Hollywood, near Birmingham, where 
several thousand pounds were expended in plant. Business 
was soon commenced on a large scale, the company produc- 
ing, however, many other alloys besides those of aluminium. 
The business grew until it soon became the serious competitor 
of the French company, and practically controlled the English 
market. However, a radical change of still greater importance 
in the sodium process was made in 1886 by an invention of 
Mr. H. Y. Castner, of New York City. This gentleman con- 
ceived the plan of reducing sodium compounds in cast-iron 
pots, from a fused bath of caustic soda, by which the reduction 
is performed at a much lower temperature, and the yield of 
sodium is very much more than by the Deville method. The 
application of this process on a large scale, with the use of gas 
furnaces and other modern improvements, has lowered the cost 
of sodium from $1 per pound to about 20 or 25 cents. It is 
but just to say that Mr. Castner's invention was by no means a 
chance discovery. For four years he worked in a large labor- 
atory fitted up for this special purpose, and after many dis- 
couragements in trying to produce aluminium by means other 
than that of sodium, was led finally to consider that the cheap- 
ening of this metal was the most promising method for cheap- 
ening aluminium, and after much patient, hard work, success 
was at last reached. 

Mr. Castner's patent was taken out in the United States in 
June, 1 886, and, while being the first one granted on that sub- 
ject in this country, is said also to be the only one taken out in 
the world since 1808. With the assistance of Messrs. J. H. and 
Henry Booth, of New York City, Mr. Castner demonstrated the 
process by building and operating a furnace on a somewhat 
large scale. This being accomplished, Mr. Castner crossed to 
England and met the representatives of the Webster process. 


with whom it was evident a combination would be especially 
advantageous to both parties ; for, with cheap aluminium chlor- 
ide and cheap sodium, it was clear that a strong process could 
be built up. Mr. Castner then demonstrated plainly, by erect- 
ing a furnace and operating it for several weeks, that his pro- 
cess was all that he claimed for it. As a result of this success, 
the "Aluminium Company, Limited," was incorporated in June, 
1887, with a share capital of .£'400,000, " to acquire the patents 
and work and develop the inventions of James Webster for the 
manufacture of pure alumina and certain metallic alloys and 
compounds, together with the business at present carried on by 
the Webster Patent Aluminium Crown Metal Company, Lim- 
ited, in Birminghami, Sheffield, and London, England, and also 
to acquire the patents and work and develop the invention of 
H. Y. Castner for the manufacture of sodium and potassium." 
Mr. Webster was paid ;£'2 30,000 for the business, properties, 
stock, etc., of the Crown Metal Company, while ;£'i40,ooo was 
allowed for the sodium patents. The new company appointed 
Mr. Castner managing director, and the erection of large works 
was immediately begun at Oldbury, near Birmingham. These 
works were started in operation at the end of July, 1888. They 
covered five acres of ground, and their daily production for 
some time was about 250 pounds, which was sold at 20 s. per 
pound. This was increased during 1889 to nearly 500 pounds 
a day, and the selling price of the aluminium decreased to 16 
shillings. Early in 1891, however, this company was forced to 
stop the manufacture of aluminium, as its selling price was re- 
duced by the owners of the electrolytic processes below 4 shil- 
lings per pound, at which price the sodium processes could no 
longer compete. The company continued in the sodium busi- 
ness, and in 1893, with their original capital of .£400,000 writ- 
ten down to ;£^8o,ooo, we hear that they are doing a prosperous 
business under Mr. Castner's direction, producing sodium and 
sodium peroxide, and making caustic soda and bleaching 
powder by electrolysis of common salt. In view of the promi- 
nent part which this works took in the aluminium industry from 


1888 to 1 89 1, and as an acknowledgment to Mr. Castner's able 
direction, the methods there used have been rightly designated 
as " The Deville-Castner Process." 

We have followed the progress of the Webster and Castner 
processes up to the date of starting the works at Oldbury, be- 
cause the continuity of the advances made in the old Deville 
process would hardly allow of a break in order to mention 
other processes arising meanwhile. However, the years since 
1884 have witnessed not one but several revolutions in the alu- 
minium industry. The great advances made in dynamo-electric 
machinery in the last decade have led to the revival of the old 
methods of electrolysis descovered by Deville and Bunsen, and 
to the invention of new methods of decomposing aluminium 
compounds electrolytically. It will be recalled that the first 
small pencils of aluminium made by Deville were obtained by 
electrolysis, and that he turned back to the use of the alkaline 
metals solely because the use of the battery to effect the de- 
composition was far too costly to be followed industrially. 
This fact still holds true, and we cannot help supposing that if 
Deville had had dynamos at his command such as we have at 
present, the time of his death might have seen the aluminium 
industry far ahead of where it now is. 

First in point of time we notice Gratzel's process, patented in 
Germany in 1883 and used industrially by the "Aluminium and 
Magnesium Fabrik, Patent Gratzel," at Hemelingen, near Bre- 
men. The process was essentially the electrolysis of a bath of a 
fused aluminium salt, such as chloride or fluoride, the improve- 
ments on the older experiments being in details of apparatus 
used, the use especially of anodes of mixed carbon and alu- 
mina, and the use of dynamic electricity. Several metallurgists 
maintained the uselessness of the Gratzel processes, and their 
position was proved to be not far from the truth, for in October, 
1887, the company announced that the addition " Pt. Gratzel" 
would be dropped from the firm name, since they had aban- 
doned Gratzel's processes and were making aluminium by 
methods devised by Herr Saarburger, director of their works. 


The processes of this latter gentleman not being published, we 
are unable to state their nature, but they were very probably 
electrolytic. In October, 1888, Mr. Saarburger reported that 
their works were producing at the rate of 12,000 kilos of alu- 
minium yearly, besides a large quantity of aluminium bronze 
and ferro-aluminium. The firm also worked up the aluminium 
and its alloys into sheet, wire, tube, etc. 

Since 1890, however, this company has retired from the alu- 
minium business, their works being now used only for the pro- 
duction of magnesium. 

A somewhat similar electrolytic process was patented by Dr. 
Ed. Kleiner, of Zurich, in 1886. Molten cryolite was decom- 
posed by two carbon poles, the heat generated by the current 
first melting the cryolite and then electrolyzing it. Since the 
motive power in this, as in all electric processes, composes one 
of the chief elements for carrying on the reduction, the Kleiner 
Gesellschaft, formed to work this method, made an attempt to 
obtain water rights at the falls of the Rhine, at Schaffhauseni 
which would furnish 15,000 horse-power. This proposition 
being refused by the government, an experimental plant was 
started at the Hope Mills, Tyddesley, Lancashire, England. 
The results were not encouraging enough to allow of commer- 
cial success, and the process was abandoned. 

An electrolytic method, different in principle from both the 
preceding, is the invention of Mr. Chas. M. Hall, of Oberlin, 
Ohio, which was patented in the United States in April, 1889, 
but applied for in 1886. The principle involved is the electric 
decomposition of alumina, dissolved in a fused bath of the 
fluorides of aluminium and other bases, the current reducing 
the dissolved alumina without affecting the solvent. This 
method is essentially different from any of the previous elec- 
trical processes, which contemplated and operated simply the 
decomposition of a molten aluminium salt, such, for instance, as 
the solvent used by Mr. Hall. The knowledge that the fused 
fluorides dissolve large quantities of alumina, and that the elec- 
tric current will act on this dissolved alumina without decom- 


posing the solvent, was the essence of Hall's invention, and the 
source of another revolution in the aluminium industry. The 
process has been operated by the Pittsburgh Reduction Com- 
pany, from 1889 to 1891, in Pittsburgh, Pa., since 1891 at New 
Kensington, Pa., on the banks of the Allegheny river, eighteen 
miles above Pittsburgh. During 1889 this company produced 
about 75 pounds of aluminium a day, which they sold at $4.50 
per pound ; in 1 890 their capacity was increased to 400 poun'ds 
per day, and the selling price reduced to $2.00 per pound; 
since 1892 the works ha,ve been again enlarged to a capacity of 
nearly one ton per day, and the selling price has tended stead- 
ily downwards until it is at present $0.50 per pound. During 
189s this company expects to put into operation a very large 
new plant at Niagara Falls, using power generated by the new 
tunnel. This works will utilize up to 6,000 horse-power, and 
will have a capacity when in full operation of 6,000 to 8,000 
pounds per day. This improvement will place the United 
States in the front as the largest aluminium-producing country 
in the world. 

In France, Adolphe Minet, a well-known electrician, experi- 
mented during 1887 and 1888 on the electrolysis of a molten 
bath containing aluminium fluoride and sodium chloride, fed 
by the addition of alumina and aluminium fluoride. From 
1888 to 1 89 1 this process was run on an experimental scale 
at the works of Bernard Bros., at Creil, making about 35 pounds * 
of aluminium per day. In 1890 a larger plant was put up, run 
by water power, at Saint Michel, Savoy, having a much larger 
capacity, which it is stated had been increased in 1894 to 
1,000 pounds per day. In 1895, this plant has been acquired 
by a French company who will operate by the Hall process. 

While the electrolytic processes so far considered use a fluid 
bath, and operate at moderate temperatures with a current of 
moderate intensity, there have been devised two other promi- 
nent processes which operate in a somewhat different manner 
and attain to economical results. These primarily depend on 
the enormous temperature attainable by the use of a powerful 


electric arc, and secondarily on the reduction of alumina (which 
at the temperature attained becomes fluid) either by the reduc- 
ing action of the carbon present or by simple electric decompo- 
sition. Which of these two agencies performs the reduction, in 
either process, is still an unsettled question, which we will dis- 
cuss later on. 

Before going further with the history of these two processes, 
Cowles' and Heroult's, it may not be inappropriate to take note 
of a few facts antecedent to their appearance. It is well known 
that Sir W. Siemens devised an electric furnace in which the 
heat of the arc was utilized for melting steel. In 1882 Mr. 
Ludwig Grabau, in Hanover, Germany, purchased a Siemens 
furnace for the express purpose of attempting the reduction of 
alumina, and after experimenting successfully for some time, 
modified the apparatus so as to work it continuously, and there- 
with made aluminium alloys ; but on account of the difficulties 
of the process, and the impurity of the alloys produced, Mr. 
Grabau gave up the experiments, having come to the conclu- 
sion that aluminium alloys to be technically valuable should be 
obtained in a state of almost chemical purity. In the beginning 
of 1885 Dr. Mierzinski, in his book on aluminium, presented 
some very striking remarks on the use of the electric furnace, 
which are so much to the point that they are well worth quoting 
in this connection : " The application of electricity for produc- 
ing metals possesses the advantage not to be ignored that a de- 
gree of heat may be attained with it such as cannot be reached 
by a blowpipe or regenerative gas-furnace. The highest furnace 
temperature attainable is 2500 to 2800° C., but long before this 
point is reached the combustion becomes so languid that the 
loss of head by radiation almost equals the production of heat 
by combustion, and hinders a further elevation of temperature. 
But in applying electricity the degree of heat attainable is theo- 
retically unlimited. A further advantage is that the smelting 
takes place in a perfectly neutral atmosphere, the whole opera- 
tion going on without much preparation, and under the eyes of 
the operator. Finally, in ordinary furnaces, the refractory 


material of the vessel must stand a higher heat than the sub- 
stance in it, whereas by smelting in an electric furnace the 
material to be fused has a higher temperature than the crucible 
itself. Since the attempt to produce aluminium by the direct 
reduction of alumina by carbon is considered by metallurgists 
as impossible, because the temperature requisite is not attain- 
able, the use of the electric current for attaining this end seems 
to be of so much the more importance." 

The Cowles invention was patented August i8, 1885, and 
was first publicly described before the American Association 
for the Advancement of Science, at their Ann Arbor meeting, 
August 28, 1885. The process is due to two Cleveland gen- 
tlemen, E. H. and A. H. Cowles, who in the development of 
their process associated with them Prof. Charles F. Mabery, of 
the Case School of Applied Science, Cleveland, as consulting 
chemist. The Cowles Electric Smelting and Aluminium Com- 
pany, formed to work the process, erected a plant at Lockport, 
N. Y., where a water power of 1200 horse-power was secured, 
and where, among other novel apparatus, the largest dynamo 
in the world, made especially for this purpose by the Brush 
Electric Company, was in operation. Following the success of 
this plant in America, the Cowles Syndicate Company, organ- 
ized to work the patents in England, put in operation works at 
Stoke-on-Trent which had a capacity of something like 300 
lbs. of alloyed aluminium daily. Springing also from the 
Cowles process is the "Aluminium Brass and Bronze Com- 
pany," of Bridgeport, Conn., which was organized in July, 
1887, and controls the exclusive rights under the Cowles 
American patents of manufacturing the alloys of aluminium 
into sheet, rods, and wire. The extensive plant of this com- 
pany employs 300 men, and has been erected at a cost of nearly 

The principle made use of in the Cowles process is, briefly, 
that a powerful electric current is interrupted, the terminals 
being large carbon rods, and the space between having been 
filled with a mixture of alumina, carbon, and the metal to be 


alloyed, the intense heat generated in contact with this mixture 
causes the metal to melt and the alumina to be reduced to alu- 
minium, which combines with the metal, while the oxygen 
escapes as carbonic oxide. 

It is interesting to note as separating the Cowles, as well as 
the Heroult, process from the previously-mentioned electrolytic 
methods, that while the latter produce almost exclusively pure 
aluminium in their electric operation, finding it inexpedient, if 
not perhaps impossible, to add other metals and form alloys at 
once — the former experience almost the reverse of these condi- 
tions, and as yet are confined exclusively to the direct produc- 
tion of the alloys. 

The Heroult process was first put in practical operation on 
July 30th, 1888, at the works of the Swiss Metallurgic Com- 
pany (Societe Metallurgique Suisse), at Neuhausen, near 
Schaffhausen. The patents for the process were granted in 
France and England in April and May, 1887, and in the United 
States in August, 1888. The company named above is com- 
posed of some of the largest metal-workers in Switzerland. 
Previously to their adoption of this process they had experi- 
mented with Dr. Kleiner's electrolytic method, but abandoned 
it, and on becoming the owners of the Heroult process imme 
diately started it up practically on a large scale, and with signal 

The process consists in electrolyzing molten alumina which 
has been rendered fluid by the heat of the arc, using as the 
positive anode a large prism of hard carbon and as the negative 
a substratum of molten copper or iron, the arrangement of the 
parts being such that the process seems to proceed, when once 
well under way, in all respects as tlfe simple electrolysis of a 
liquid. Using water power for driving the dynamos, the 
economical production of alloyed aluminium at 4.5 francs per 
kilo (50 cents per pound), was an assured fact. 

The success of this process at Neuhausen was so marked as 
to attract general attention, and in the latter months of 1 888 
several large German corporations, prominent among which was 


the Allgemeine Electricitats Gesellschaft of Berlin, sent repre- 
sentatives to arrange for the purchase of the Heroult patents 
for Germany. The outcome of these examinations and negotia- 
tions was the purchase by this German Syndicate of HerouU's 
continental patents and the founding by them and the former 
Swiss owners of the Aluminium Industrie Actien-Gesellschaft, 
with a capital of 10,000,000 francs. In December, 1888, the 
new company took possession at Neuhausen, and commenced 
the construction of a plant many times larger than the original 
one, their plans also including the erection of foundries and 
mills for casting and manufacturing their alloys. Dr. Kiliani, 
the well-known writer on electro-metallurgical subjects, was 
working manager for the company.* The new plant utilized 
3000 horse power, and had a capacity of 3000 pounds of alloyed 
aluminium daily. 

In the same year, the Societe Electro-Metallurgique Fran- 
caise established a works at Froges (Isere) 12 miles from 
Grenoble, to work the Heroult process, with Kiliani's modifica- 
tions. They used 400 horse-power, which later was enlarged 
to 800, and in 1894 acquired a large water power at La Praz, 
near Modane, where works have been erected and lately put 
into operation. 

The process now used here and in Switzerland is not the 
original Heroult process as previously described. Dr. Kiliani 
introduced some modifications in the apparatus, but no pure ' 
aluminum was made by either of the Heroult works until the 
end of 1889, when there was begun the use of cryolite in the 
furnace to reduce the melting point of the alumina and the 
electrical tension required to operate it. As thus modified, the 
process is essentially identical in principle with Hall's process ; 
in fact, the U. S. Patent office refused Heroult's application for 
a patent on account of Hall's priority to the process. An at- 
tempt was made in 1889 to work the Heroult alloy process in 
the United States, an experimental plant being started at 

* Dr. Kiliani died suddenly in January, 1895, in the midst of his duties. 


Bridgeport, Conn., in August of that year, but it was stopped 
by the burning out of the dynamo. In 1890 another dynamo 
arrived from the Oerlikon works, at Zurich, and the process 
was operated for some time on an experimental scale at Boon- 
ton, N. J., but was abandoned in 1891. 

During the years 1885 to 1890 both the Cowles and Heroult 
alloy processes were successful in producing aluminium in al- 
loys at a lower cost than the market price of pure aluminium, 
and therefore built up a considerable business. Since pure 
aluminium has reached so low a price, metal mixers prefer to 
make their own alloys from the pure metal, and the ready-made 
alloys have lost their market. 

The Alliance Aluminium Company of London, England, was 
organized in the early part of 1888, with a nominal capital of 
^500,000. This company owned the English, German, French, 
and Belgian patents of Prof. Netto, of Dresden, for the manu- 
facture of sodium and potassium and the reduction of cryolite 
thereby ; the patents of Mr. Cunningham for methods of reduc- 
tion of the same metals ; and methods devised by Prof. Netto 
and Dr. Saloman, of Essen, for producing aluminium of great 
purity on a commercial scale. The two latter named gentle- 
men are said to have invented their processes after long ex- 
perimenting at Krupp's works at Essen; and, since the ap- 
paratus used was mounted on trunnions, many rumors were 
spread by the newspapers that aluminum was being made (by 
tons, of course) in a Bessemer converter by Krupp, of Essen. 
Prof. Netto reduced sodium by a continuous process, by allow- 
ing fused caustic soda to trickle over incandescent charcoal in a 
vertical retort, the apparatus containing many ingenious details 
and giving promise of being quite economical. One method of 
using the sodium in reduction consisted in the use of a plunger 
to which bars of sodium were attached and held at the bottom 
of a crucible full of molten cryolite ; another depended on the 
use of a revolving cylinder in which the cryolite and sodium 
reacted, and appears more chimerical than Netto's other pro- 
positions. This latter device, however, is said to have been 
operated at Essen. 


In June, 1888, the Alliance Company were located at King's 
Head Yard, London, E. C, and several small reduction furnaces 
were being operated, each producing about 50 lbs. of alumin- 
ium a day, estimates of the cost at which it was made giving 
6 to 8 shillings per pound. In the early part of 1889 the 
" Alkali Reduction Syndicate, Limited," leased ground at Hep- 
burn on which to erect works for making sodium by Cunning- 
ham's patents, the sodium produced being sent to Wallsend, 
near Newcastle-on-Tyne, where the Alliance Company's reduc- 
tion works were located. 

In 1890 this company could not profitably compete with the 
electrolytic processes, and in 1 892 their works were sold at auc- 

Ludwig Grabau, of Hanover, Germany, has made several 
patented improvements in producing aluminium, which are in 
the same direction as Prof. Netto's methods. Mr. Grabau be- 
lieves that in order that aluminium may possess its most valuable 
qualities, either for use alone or in alloying, it should be of 
almost chemical purity, and as the best means of attaining this 
end economically he has improved the sodium method on 
these three lines : — 

1st. Production of cheap pure aluminium fluoride. 

2d. Production of cheap sodium. 

3d. Reduction in such a manner that no possible impurities 
can enter the reduced metal, and that the sodium is completely 

Mr. Grabau has devised processes for making pure alumin- 
ium fluoride from kaolin, sodium by electrolysis of fused salt, 
and has succeeded in producing some very pure aluminium, 
but not cheaply enough to compete with the Hall and Heroult 

Col. William Frishmuth, of Philadelphia, operated a small 
chemical works, and before i860 made sodium by Deville's 
process, and with it made aluminium in small quantities. He 
claimed to have other methods of producing aluminium cheaply, 
but none of them were practicable. He deserves credit for hav- 


ing made a good casting of the aluminium tip on tlie Wasiiing- 
ing Monument at our national capital. 

In 1889 the aluminium industry was finely represented at the 
Paris Exposition. It was the period in which the sodium pro- 
cesses were at their culmination, and the electrical processes 
just becoming serious competitors. A detailed account of the 
exhibits reads as follows : 

Societe Anonyme pour I'lndustrie de 1' Aluminium : In a 
large case, the frame of which was aluminium bronze, samples 
of aluminium, ferro-aluminium, aluminium bronze, forged and 
rolled, and numerous articles of the latter alloy. Cowles Elec- 
tric Smelting and Aluminium Company : Samples of ferro-alu- 
uminium, aluminium bronze and aluminium brass of various 
grades, aluminium silver, and numerous useful articles made of 
these alloys. Brin Bros. : Samples of aluminium with thin iron 
and steel castings made by its use. The Alliance Aluminium 
Company:. Two large blocks of aluminium, cast hollow, weigh- 
ing possibly 1,000 pounds and 500 pounds, respectively. The 
inclosing balustrade and decorations were principally of alu- 
minium or aluminium bronze. The Aluminium Company, 
Limited : A solid casting of aluminium bronze weighing y^ ton, 
and on this a solid block of 98 per cent, aluminium weighing 
the same. In the corners of the case piles of ingots of 99 per 
cent, aluminium, 10 per cent, bronze, 5 per cent, bronze, 10 
per cent, ferro-aluminium, and 20 per cent, aluminium steel. 
Besides which were a 7-inch bell, springs, statues, aluminium 
plate, round and square tubes, wire, sheet, etc. Such were the 
aluminium exhibits, which attracted as much interest as the 
historic ingot of 1855 did at its debut, and, not taking into ac- 
count that Mr. Hall's process was not represented and that the 
German makers were debarred from exhibiting because of in- 
ternational pique, yet the exhibit shown was one which demon- 
strated the great advances made in the previous five years. 

Since 1 890, the progress of the aluminium industry has been 
steady but sure. The sodium processes have dropped alto- 
gether from the race, and the electrical processes occupy the 


field. Among these, the processes producing alloys only have 
dropped out of sight. Among those that remain, only those 
favorably situated, near abundant water power and with cheap 
supplies of alumina, have flourished. The industry has reached 
a strict commercial basis, in which every cent of cost has to be 
considered, and every item of expense reduced to a minimum. 

In 1890 the Cowles Electric Smelting and Aluminium Com- 
pany began selling pure aluminium. They were sued by the 
Pittsburgh Reduction Company for infringement of the Hall 
patents, and in the early part of 1893 were enjoined by the 
courts from continuing their pure-metal process, it being de- 
clared an infringement of Hall's patents. 

Mention should be made of the aluminium exhibit at the 
Columbian Exposition in Chicago, 1893. The only makers 
represented were the Pittsburgh Reduction Company, but their 
exhibit was a very satisfactory representation of the state of the 
art. A working model of an extraction pot was shown, hun- 
dreds of pounds of the metal in all shapes and worked in every 
way, a fine set of aluminium alloys, and several cases of manu- 
factured goods made from their metal by various American 
firms. Mr. Hirsh of Chicago showed aluminium electroplated 
on other metals and on wood. The writer exhibited a metal- 
lurgical suite showing the minerals from which aluminium is 
extracted, samples of aluminium made by the various processes 
which have been used, samples showing the physical properties 
of the metal, its working, soldering, etc., and a large number of 
its alloys. 

The statistics about to be given will show graphically the 
story of the rise of the aluminium industry. The last ten years 
have witnessed an undreamt-of progress. Hereafter the lower- 
ing in price can only be counted at a few cents a pound, but 
the increase of production may be noted in thousands of tons. 
A most fitting ending to this historical sketch would be simply 
to repeat what I wrote three years ago, viz., " The nineteenth 
century will live in history as that century which gave to the 
world the railway, the telegraph, the telephone, the dynamo, 
Bessemer steel and aluminium." 



The following table shows the price at which aluminium has 
been sold since it was first placed on the market : 

Dale. Place. Per kilo. Per pound. 

1856 (Spring) Paris looo fr. ^90.90 

1856 (August) " 300 " 27.27 

1859 " 200" 17-27 

1862 " 130" 11.75 

1862 Newcastle ii-75 

1878 Paris 130" 11.75 

1886 " 12.00 

1887 Bremen 8.00 

1888 London 4.84 

1889 Pittsburgh 2.00 

1891 " 1 1.50 

1892 " i.oo 

1893 " 0-75 

1894 " 0.50 

1895 Switzerland 3 M. 0.35 

Until 1885 the selling price of aluminium bronze depended 
directly on the price of aluminium, being about $1.50 per 
pound for the ten' per cent, alloy. With the advent of the 
Cowles and, later, the Heroult alloy-processes, the price fell 
below what the contained aluminium would have cost, as is 
shown by the selling price of the ten per cent, bronze, as fol- 

Date. Place. Per kilo. Per pound. 

1878 Paris 18.00 fr. gi.64 

1885 Cowles Bros 4.50 " 0.40 

1888 " " 3-85 " 0.35 

1888 Heroult process, Neuhausen 3.30 " 0.30 

Since 1889 the cost of the bronze has again been dependent 
on the cost of the pure metal, so that no separate figures are 
necessary, With copper at $0.10 per pound, and aluminium 
at $0.50, the cost of 10 per cent, bronze is not over $0.20 per 

As to the amount of aluminium which has been produced. 


we can make the following estimates, gleaned from various 
sources : 



1854-56 Deville 25 

1859 Nanterre (Deville) 720 

1859 Rouen (Tissier Bros.) 960 

1865 France 1,090 

1869 " 455 

1872 Sallndres (H. Merle & Co.) 1,800 

1882 " " " 2,350 

1884 " " " 2,400 

1887 France 3,500 

1888 " 4,500 

1889 " 14,840 

1890 " 37,°oo 

1891 " 36,000 

1892 " 40,000 

A careful estimate of the total amount made in France by 
the Deville process, from 1855 to 1888, places it at 43,000 
kilos. Since 1888, there has been made by other processes 
128,000 kilos up to the year 1893, making a total of 171,000 
kilos, or 375,000 pounds. 



1872 Bell Bros 1,650 

1889 Castner process 50,000 

1891 Electrolytic processes 90,000 

1892 " " 90,000 

The total English production up to 1890 was probably not 
over 70,000 pounds. The Castner process produced altogether 
from July, 1888 to 1891, 250,000 pounds. The total produced 
in England to the end of 1 892 is therefore in the neighborhood 
of 450,000 pounds. 



1890 Neuhausen 40,540 

1891 " 168,670 

1892 " 300,000 

1893 " 480,000 

1894 •' 600,000 


The total production of these works to the end of 1894 was 
over 1,500,000 kilos, or 3,300,000 pounds. 

United States. 


1883 Frishmuth 70 

1884 " 125 

1885 " 230 

1885 Cowles Bros, (in alloys) 450 

'886 " " 6,500 

•887 " " 17,800 

1888 Pure and alloyed 19,000 

1889 " 47,468 

1890 " 61,281 

1891 " 150,000 

1892 " 259,885 

1893 " • 333,629 

1894 '• 706,000 

World's Production to End of 1892. 

Kilos. Pounds. 

France 171,000 376,000 

England 204,500 450,000 

Switzerland 500,000 1,100,000 

Germany (1885- 1 889) 50,000 110,000 

United States 250,000 550,000 

Total 1,175,500 2,586,000 

The amount made in 1 893 was as follows : 

France (estimated) 40,000 kilos. 

Switzerland 480,000 " 

United States 150,000 " 

Total 670,000 " =: 1,474,000 pounds. 

An estimate for 1 894 would be as follows : 

France 100,000 kilos. 

Switzerland 600,000 " 

United States 320,000 " 

Total 1,020,000 " = 2,244,000 pounds. 

The total amount of aluminium made from 1855 to the be- 


ginning of 1895 is, therefore, in the neighborhood of 2,865,500 
kilos, or 6,304,000 pounds, while the annual production has 
reached 1,000 long tons, and will probably be increased during 
1895 to nearly 2,000 tons. 

In comparing these figures with the production of other 
metals, it is important to bear in mind the low specific gravity 
of aluminium. For most practical applications, the bulk of the 
metal used is a fixed quantity, and 1,000 tons of aluminium 
are equal in volume to 

4,600 tons of lead. 
4,000 " " silver. 
3,500 " " nickel. 
3,300 " " copper. 
2,800 " " tin. 
2,700 " " zinc. 



There is no other useful metal, iron not excepted, which 
is so widely scattered over the earth and occurs in such 
abundance. F. W. Clarke of the Smithsonian Institution has 
calculated that the percentage of aluminium in the earth's crust, 
as far as it is known to us, is 7.81, while that of iron, the next 
most abundant element, is 5.46. 

Aluminium is not found metallic. Stocker * made the state- 
ment that aluminium occurred as shining scales in an alumina 
formation at St. Austel, near Cornwall, but he was in error. 
But the combinations of aluminium with oxygen, the alkalies, 
fluorine, silicon, and the acids, etc., are so numerous and occur 
so abundantly as not only to form mountain masses, but to be 
also the bases of soils and clays. Especially numerous are the 
combinations with silicon and other bases, which, in the form 
of feldspar and mica, mixed with quartz, form granite. 

These combinations, by the influence of the atmospheric air 
and water, are decomposed, the alkali is replaced or carried 
away, and the residues form clays. The clays form soils, and 
thus the surface of the earth becomes porous to water and 
fruitful. It is a curious fact that aluminium is rarely found in 
animals or plants. According to Church, f alumina is found 
only in a few vascular cryptogamous plants, whose ash con- 
tains up to 20 per cent. 

Ricciardi concluded from numerous experiments that the 
assimilation of alumina by plants does not depend upon the 
abundance of alumina in the soil, and that, generally speaking, 

* Joum. fr. prakt. Chem., 66, 470. 

t Proc. Royal Society, Vol. 44, p. 121. 



there is more found in the trunk and branches, less in the 
husks and seeds, and least in the leaves. 

Most of the aluminium compounds appear dull and disagree- 
able, such as feldspar, mica, pigments, gneiss, amphibole, por- 
phyry, eurite, trachyte, etc. ; yet there are others possessing 
extraordinary lustre, and so beautiful as to be classed as prec- 
ious stones. Some of these, with their formulae, are : 

Ruby AI2O3 

Sapphire Al^Oj 

Garnet (Ca.Mg.Fe.Mn)3Al2Si30,2 

Cyanite AUSiOj 

Some other compounds occurring frequently are : 

Turquoise AljPPs.HjAljOs.aHjO 

Lazulite (MgFe)AljP203 -|- Aq 

Wavellite 2AI2P2O8.H1.AI2O6.9H2O 

Topaz 5AL,Si05.Al,SiFi„ 

Cryolite Al^Fg.eNaF 

Diaspore AIJO3.H2O 

Bauxite AI2O3.2H2O 

Aluminite AlaSOe.gH^O 

Alunite K^s6i.Al^Sfi,^.2H^Alfii 

Soda Feldspar NajAl2Si60ig 

Potash Feldspar KjAljSigOjj 

Lime Feldspar CaAljSijOg 

Kaolin Al2SijO,.2H20 

Bauxite and cryolite are the minerals most used for produc- 
ing aluminium, and their preference lies mainly in their purity. 
Native alums generally occur in out of-the-way places and not 
in beds of very large extent. 


Bauxite is a hydrous aluminium oxide of variable composi- 
tion, sometimes approaching diaspore in its proportion of water, 
but with the aluminium always more or less replaced by iron, 
and some silica disseminated through it. Its color is creamy 
white when free from iron, and its structure usually pisolitic, 
that is, in small globules like peas. It was first found in 


France, near the town of Baux, large deposits occurring in 
the departments of Var and Bouches du Rhon, extending from 
Tarascon to Antibes. Several of these beds are a dozen yards 
thick, and sixteen kilometers in length. Deposits are also 
found in the departments of I'Herault and I'Arriege. Very im- 
portant beds are found in Styria, at Wochein, and at Freis- 
stritz, in Austria, a newly discovered locality where the mineral 
is called Wocheinite. Here it has a dense, earthy structure, 
while that of France is conglomerate or oolitic. Deposits 
similar to those of France are found in Ireland at Irish Hill, 
Straid, and Glenravel. Further deposits are found in Hadamar 
in Hesse, at Klein Steinheim, Langsdorff, and in French 

In the United States large beds have been found in Alabama, 
Georgia, and Arkansas. Those of Georgia have been worked 
since 1890, and are now affording a large part of all the 
bauxite consumed in this country. The deposits of the Coosa 
Valley in Georgia and Alabama are worked by three com- 
panies, the Republic Mining and Manufacturing Company, the 
Georgia Bauxite and Mining Company, and the Southern 
Bauxite Mining and Manufacturing Company. These three 
companies produced, in 1893,9,200 tons, which was three-fifths 
of the total bauxite consumed in the United States, the rest 
being imported from Ireland. The ore is found in beds or 
pockets, and while the quantity in sight is not inexhaustible, 
yet there is sufficient for many years to come, at the pres- 
ent rate of mining. These bauxites are as pure as those 
of Europe, and the manufacturers using them say they are 
more easily worked and so prefer them to the imported, even 
though they often carry less alumina and cost more. The 
Arkansas ore occurs in tertiary formations, in irregular de- 
posits of 5 to 40 feet in thickness, not over 300 feet above tide 
water. H. L. Fletcher, of Little Rock, owns much of the land 
on which they occur. 

The cost of imported bauxite is $5 to $7 a ton ; of Ameri- 
can bauxite $5 to 12 per long ton, according to where the 



market is; best selected Georgia bauxite brings $io per ton 
in New York city. 

The following are analyses of foreign bauxites ; besides the 
ingredients given there are also traces of lime, magnesia, sul- 
phuric, phosphoric, titanic and vanadic acids. 

Alumina ■ • ■ 
Ferric oxide 


Alkalies. • . . 



























Alumina ■ ■ . 
Ferric oxide 


Alkalies . ■ - 

Alumina • • • 
Ferric oxide 














2.1 1 


















1 1.0 












6.4 1 





Index: — 

I and 2. From Baux (Deville). 

Dark \ 

T ■ vj ]■ Wocheinite (Drechsler). 

Red brown \ 

Yellow >■ Bauxite from Freisstritz (Schnitzer). 

White > 

White Wocheinite (L. Mayer and O. Wagner). 

Bauxite from Irish Hill. 



10. Bauxite from Co. Antrim (Spruce). 

11. « " Glenravel (F. Hodges). 

12 and 13. " " Hadamar in Hesse (Retzlaff). 

14. From Klein-Steinheim (Bischof). 

15 and 16. From Langsdorfi (I. Lang). 

17. Bauxite from Dublin, Ireland, brought to the Laurel Hill Chemical 

Works, Brooklyn, L. I., and there used for making alums. It is dirty 
white, hard, dense, compact, and in addition to the ingredients given 
above contains 0.59 per cent, of lime and some titanic acid. It costs 
$6 per ton laid down in the works. The above analysis, made by Mr. 
Joiiet, is furnished me by the kindness of the superintendent of the 
works, Mr. Herreshoff. 

The American bauxites are fully equal to the foreign in 
quality, as is shown by the following analyses : 

Ferric Oxide. 



Titanic Acid. 


Alumina | 37-62 

























Alumina — 
Ferric Oxide 



Titanic Acid 



















18. Little Rock Region, Pulaski Co., Arkansas. 

jg >I « " " " 

20. Red variety. Saline Co., Arkansas. 

21. Pink " 

22. White variety, Floyd Co., Georgia. 


24. Georgia Bauxite Company. 
2c " " " 

26. Wharwhoop Mine, Cherokee Co., Alabama. 

27. Red variety, Cherokee Co., Alabama. 


Regarding French bauxites, Francis Laur, the first to work 
the French deposits, states* that one of their peculiarities is 
that the bulk of that mined contains a nearly constant amount 
of alumina, of silica, ferric oxide and water taken together, and 
of other impurities. The proportions are very nearly 

Alumina 68 to 70 per cent. 

Silica, ferric oxide and water 27 " 

Other impurities 3 to 4 " 

In the bauxite from Baux, that first worked, the silica, ferric 
oxide and water are present in about equal proportions ; that 
is, about 8 to 10 per cent, of each. In the pale bauxite from 
Villeveyrac (Herault) there is almost no iron, and the 27 per 
cent, is nearly evenly divided between silica and water ; what 
small amount of ferric oxide does occur in it can be put into 
the list of " other impurities." The red bauxite, discovered in 
1880 in the south of France, is very constant in composition, 
and may be regarded as the inverse of the pale variety. In this 
red kind the silica is so small as to be relegated to the acces- 
sory constituents, while water and ferric oxide alone make up 
the 27 per cent. These three varieties represent the three types 
of bauxite which are mined in large quantities. The pale 
variety, being low in iron, is preferred for making aluminium 
sulphate by treatment with sulphuric acid, which attacks it very 
readily; the red kind is preferred for the sodium carbonate 
treatment, as the iron does not interfere, while silica is kept out 
of the alumina produced. Besides these kinds, some varieties 
are found with the 27 per cent, either all water, all silica or all 
ferric oxide. We would thus have, as the composition of the 
typical kinds of bauxite found in France, the following: 

Water. Silica. Ferric oxide. 

Hyaline bauxite 27 o o 

Silicious type, compact o 27 o 

Ferruginous type, in pisolites o o 27 

Red bauxite of Var 131^ o 131^ 

Pale bauxite of Villeveyrac 131^ i^}4 o 

Mixed bauxite of Baux 9 9 9 

* Trans. Am. Inst. Mining Eng., Feb., 1894. 


Mr. A. E. Hunt, of Pittsburgh, in the discussion of this paper, 
remarked that there are vast quantities of bauxite in Georgia, 
Alabama and Arkansas, which run low in both silica and ferric 
oxide, containing less than 5 per cent, of both these ingredients, 
but with about 4.5 per cent, of titanic acid. It is his experience 
that when the water of hydration is below 31 per cent., in the 
greyish and white bauxite from Georgia and Alabama, the 
silica is always high. If the bauxite is red, however, the water 
may be low without the silica being high. He cites the follow- 
ing analyses illustrating this point : 

Combined water. 


Ferric oxide. 

Titanic acid. 



























It will be seen that Mr. Laur's generalizations regarding 
French bauxites do not hold good for the American varie- 
ties ; the constancy of the titanic aci,d is also remarkable. 


Cryolite was first found at Ivigtuk in Arksut-fiord, west coast 
of Greenland, where it constitutes a large bed or vein in gneiss. 
It was very rare even in mineralogical collections until 1855, 
when several tons were carried to Copenhagen and sold under 
the name of " soda mineral." It is a semi-transparent, snow- 
white mineral. When impure it is yellowish or reddish, even 
sometimes almost black. It is shining, sp. gr. 2.95, and hard- 
ness 2.5 to 3. It is brittle, not infrequently contains ferrous 
carbonate, sulphide of lead, silica, and sometimes columbite. 
It is fusible in the flame of a candle, and on treatment with 
sulphuric acid yields hydrofluoric acid. As will be seen further 
on, cryolite was first used by the soap-makers for its soda ; it 
is still used for making soda and alumina salts, and to make a 
white glass which is a very good imitation of porcelain. The 
Pennsylvania Salt Company in Philadelphia import it from 


Ivigtuk by the ship-load for these purposes ; lately they have 
discontinued making the glass. Cryolite is in general use as a 
flux. A very complete description of the deposit at Ivigtuk 
can be found in Hoffman's " Chemische Industrie." 
Pure cryolite contains 

Aluminium l3-° 

Fluorine 54-5 

Sodium 32.5 


Or otherwise stated, 

Aluminium fluoride 40.25 

Sodium fluoride ■* 59-75 


From the reports in the Mineral Resources of the United 
States we find that there was imported by the Pennsylvania 
Salt Company in 1892, 8000 long tons, which was valued at 
$9 a ton. The importers say this value is too low ; they sell 
what they call pure prepared cryolite at $60 a ton. This so 
called pure article was found by Prof. Rogers, of Milwaukee, to 
contain two per cent, of silica and one per cent, of iron ; in 
fact, the whitest cryolite always contains silica microscopically 
disseminated through it, which cannot be separated out by any 

The only known deposit of cryolite in the United States is 
that found near Pike's Peak, Colorado, and described by W. 
Cross and W. F. Hillebrand in the "American Journal of 
Science," October, 1883. It is purely of mineralogical import- 
ance and interest, occurring in small masses as a subordinate 
constituent in certain quartz and feldspar veins in a country 
rock of coarse reddish granite. In composition, however, it is 
practically identical with the Greenland cryolite, so that the 
following analysis by Mr. Hillebrand will stand for a typical 
analysis of cryolite as found in nature : 


Fluorine 53.55 per cent. 

Aluminium , 12,81 " 

Sodium 32.40 " 

Ferric Oxide 0.40 " 

Lime 0.28 " 

Water 0.30 " 


Until 1869 the sole sources of corundum were a few river 
washings in India and elsewhere, where it was found in scat- 
tered crystals. Its cost was twelve to twenty-five cents a 
pound. Within the last twenty years numerous mines have 
been opened in the eastern United States, the first discovery of 
which was due to Mr. W. P. Thompson, and is thus described 
by him : * 

" In 1869, in riding over a spur of the Alleghenies in north- 
ern Georgia, I found what has proven to be an almost inex- 
haustible mine of corundum in the chrysolite serpentine, the first 
instance on record of the mineral being found. /« situ. Previ- 
ously it had been washed out of debris at Cripp's Hill, N. C, 
and at a mine in West Chester, Pa., both on the slopes of the 
chrysolite serpentine. The clue being thus obtained accident- 
ally, about thirty mines were shortly afterwards discovered in 
the same formation ; but of the thousands of tons thus far dug 
out, the larger portion has come from the mines I discovered. 

" At present it can be bought at about ten dollars per ton at 
the mines. It is nearly pure alumina. Disapore, a hydrated 
alumina, is also found in the same region and locality. Corun- 
dum will probably always be the principal source in America 
of material from which to manufacture pure aluminium ; but in 
Great Britain, in all probability, manufacturers must look to 
alumina prepared artificially from cryolite or from sulphate of 

In 1892 the production of corundum in the United States 
was somewhat over 2000 short tons, valued at $40 a ton when 
sifted and broken to proper sizes for the market, the Hamp- 

* Journal of the Society of Chemical Induustry, April, 1886. 


den Emery Company, at Laurel Creek, Ga., and at Corundum 
Hill, N. C, and the Unionville Corundum Mines Company, at 
Unionville, Chester Co., Pa., being the principal producers. 
Thus, while corundum may be worth in bulk at the mines only 
$io a ton, yet it is extremely hard to crush, and when broken 
up fine is worth far more as an abrasive than as an aluminium 
ore. The present large development of bauxite mining has 
completely nullified the use of corundum as a source of alumin- 
ium, and I do not suppose that at present a single pound of 
aluminium is being made from it. 


The mineral kaolin is variously known as white china clay, 
fuller's- earth, or pure white clay. It is essentially a hydrated 
silicate of aluminium, containing when pure 

Alumina 39.8 per cent. 

Silica 46.3 

Water 13.9 " 

Its chemical formula is Al2O3.2SiO2.2H2O. If the water is 
driven off by calcination, the residue contains 46.2 per cent. 
of alumina, or 24.4 per cent, of aluminium. 

Kaolin has been formed mostly by the atmospheric decom- 
position of feldspar, the alkalies of which are changed into 
carbonates by the carbonic acid of the rain water, and are then 
leached out, carrying with them part of the silica in solution, 
and leaving kaolin, according to the reaction, 


Kaolin acquires a certain plasticity when mixed with water. 
Hydrochloric and nitric acids have no action on it, but cold 
sulphuric acid dissolves its alumina, setting the silica at liberty. 
It is infusible unless contaminated with particles of feldspar or 
calcium sulphate, carbonate, or phosphate. The specific gravity 
of kaolin is 2.3. If fused with six times its weight of caustic 
potash, the resulting mass gives up potassium aluminate when 


washed with water, An analogous result is obtained with 
sodium carbonate. Grabau has prepared very pure aluminium 
fluoride from kaolin by the use of sulphuric acid and subse- 
quent boiling with fluorspar. 

Large deposits of kaolin are found in many parts of the 
world. The best locality in Europe is near Limoges, France, 
with which the famous Sevres porcelain is made. Large beds 
are found in Cornwall and West Devon, England. In the 
United States the largest deposits are at Wilmington, Del., and 
in South Carolina, Georgia and Alabama. These deposits are 
worked extensively, and the product used by the pottery, 
paper, cotton, and various other industries. The value at the 
mines is very small, it not costing over $0.50 per ton to mine 
it, but for many purposes it must be crushed, washed, and even 
floated, which increases its value to $5 or $10 a ton in New 
York city. 

The purity of kaolin, and the almost inexhaustible beds in 
which it occurs, stamp it as the natural ore of aluminium. All 
that is needed is a cheap process for separating the pure alu- 
mina from the silica, in which case no other mineral could 
compete with it as the main staple of the aluminium industry. 
This is a promising field for chemical experimenters, for such 
a process would be of great value. 

Common clays are either very impure kaolin, or else they 
contain chemically a much larger proportion of silica. The 
percentage of silica may range from 50 per cent, up to 70 per 
cent, while that of the alumina present ranges downwards from 
35 to 15 per cent. These common clays are far too impure 
and low in silica to serve as a source of aluminium, and in no 
case could they compete with kaolin. 

Some anhydrous silicates of aluminium are found in nature, 
such as Disthene, Andalusite and Cyanite, whose formula is 
AUOj.SiOj, containing 62.5 per cent, of alumina ; but they do 
not occur in large quantities, and, even if they did, are so hard 
and difficult to decompose that they would never be preferred 
to kaolin. 

50 aluminium. 

Native Sulphate of Alumina. 

In the summer of 1884, a large deposit of rock called " native 
alum" was discovered on the Gila River, Socorro County, New 
Mexico, about two miles below the fork of the Little Gila, and 
four miles below the Gila Hot Springs. The deposit is said to 
extend over an area one mile square and to be very thick in 
places. The greater part of the mineral is impure, as is usual 
with native occurrences, but it is thought that large quantities 
are available. A company formed in Socorro has taken up the 
alum-bearing ground. Through the kindness of Mr. W. B. 
Spear, of Philadelphia, the author was enabled to get a speci- 
men of the mineral. 

It is white, with a yellowish tinge. On close examination it 
is seen to consist of layers of white, pure-looking material ar- 
ranged with a fibrous appearance at right angles to the lamina- 
tion. These layers are about one-quarter of an inch thick. 
Separating them are thin layers of a material which is deeper 
yellow, harder and more compact. The whole lump breaks 
easily, and has a strong alum taste. On investigation, the 
fibrous material was found to be hydrated sulphate of alumina, 
the harder material sulphate of lime. Analysis showed 7 to 8 
per cent, of insoluble material, the remainder corresponding to 
the formula Al2(S04)3.i8H20. A small amount of iron was 

State Geologist Waring informs the writer that the deposits 
occur only in sheltered spots protected from the rain, that the 
region is eruptive, and that where the trachyte rock is being 
decomposed by sulphuric gases and waters, the salts rise to the 
surface and form crusts in protected places. Often the earth 
contains 10 per cent, or more of the anhydrous sulphate. 

A similar mineral occurs in Purgatory Valley, 12 miles east of 
Trinidad, Colorado. 

A variety of the same mineral containing some silica was 
found in a stratum 25 feet thick by the Congo Coal Company, 
while sinking a shaft at Congo, Ohio, in 1893. An analysis of 
this by Dr. Lisle, of Springfield, Ohio, gave 


Alumina 50.16 per cent. 

Silica 28.98 " 

Lime 0.46 " 

Sulphuric acid 16.21 " 

Water 4.28 " 

These minerals would appear to ofifer a cheap source of 
alumina, as the operations necessary for treating them are very- 
simple, being solution in warm water, filtration, evaporation, 
and roasting. The excessive freight charges are against the 
New Mexico and Colorado deposits, but that in Ohio should 
be of considerable economic importance if the reports concern- 
ing its extent are verified. 



Commercial aluminium is never chemically pure, and there- 
fore displays properties varying more or less from those of the 
pure .metal according to the character and amount of impurities 
present. In this treatise, whenever the properties of aluminium 
are mentioned, they must be understood to refer to the chemi- 
cally pure metal, and not to the commercial article, unless spe- 
cifically stated. 

The impurities most frequently present in commercial alu- 
minium are iron and silicon. These are found in all brands, 
varying in amount from i per cent, in the purest to 6 and even 
8 per cent, in the worst. Besides these, various other impuri- 
ties are found coming from accidental sources in the manufac- 
ture ; thus, some of the first metal made by Deville contained 
a large amount of copper (Analysis i), coming from boats of 
that metal which he used in his experiments. Metal made 
later by Deville contained zinc, coming from zinc muffles which 
he had borrowed and used for retorts, old retorts broken up 
having been used in the composition of the new ones. More 
recently, aluminium has been produced by the agency of 
sodium in the presence of lead, which latter it takes up in small 
amount. Sodium is liable to remain alloyed in very small pro- 
portion, yet it is an element so easily attacked that it destroys 
some of the most valuable qualities of the aluminium. The 
distinct effect, however, of each of these usual impurities in 
modifying the physical properties of aluminium has not yet 
been investigated in a thoroughly satisfactory manner. A few 
years more, however, of increasing familiarity with and hand- 
ling of the metal on a large commercial scale will, I believe, 



cause the effect of foreign elements on aluminium to be as 
plainly recognized as is now the case with carbon and the 
metalloids in iron. In general, we may say that silicon seems 
to play a role in aluminium closely analogous to that of carbon 
in iron ; the purest aluminium is fibrous and tough, but a small 
percentage of silicon makes it crystalline and brittle. Further- 
more, a considerable proportion occurs in the graphitoidal 
state, and may be separately estimated as graphitoidal silicon 
by dissolving the metal in dilute hydrochloric acid in the pres- 
ence of bromine water. A small amount of silicon, however,, 
gives pure aluminium a bluish cast, a disagreeable smell, and 
greatly lessens its resistance to corroding agents. A small 
proportion of iron makes the metal denser and stronger, but a 
quantity above i per cent, quickly makes it brittle. A small 
quantity of copper whitens aluminium, makes it much stronger 
and lessens its shrinkage in casting, but with over 5 per cent, 
it becomes brittle. 

As aluminium was made by the sodium processes, it con- 
tained as impurity about equal quantities of iron and silicon ; as 
made by the electrolytic processes, the iron present is usually 
much less than the silicon. Carbon can only exist in alumin- 
ium that has been subjected to an intense heat, such as in an 
electric furnace, when it may be absorbed in small quantity, 
and makes the metal brittle ; it does not occur, however, in 
ordinary commercial aluminium. Titanium may occur in 
aluminium made from raw bauxite, which frequently contains 
titanic oxide ; its presence in small amount makes the metal 
much stronger, but gives it a bluish tint. There is needed a 
systematic investigation of the effect of small amounts of impuri- 
ties on the physical and chemical properties of pure aluminium. 

The following analyses are of considerable historic interest as 
showing incidentally the improvement which has been made in 
the quality of the commercial metal since the beginning of the 
industry : 




1. Deville Process 88.350 

2. " " 92.500 

3. " " 92.000 

4. " " 92.969 

5. " " 94-700 

6. " " 96.160 

7. Tissier Bros 94.800 

8. Morin & Co., Nanterre 97.200 

9. " " 97.000 

10. " " 98.290 

11. " " 97.680 

12. Merle & Co., Salindres 96.253 

13. " " 96.890 

14. " " 97-400 

15. " " 97.600 

16. Frishmuth 97-49 

17- " 97-75 

18. Hall's Process 98-34 

19. Deville-Castner - . 99.20 

20. Grabau Process 99.62 

21. " " 99-So 

22. Hall's Process 99-93 


























. 3-293 



1. 00 


















Notes on the above analyses :- 











12, 13. 

14. 15- 


Analyzed by Salvetat. Contained also 6.38 per cent, of copper and a trace 

of lead. 
Analyzed by Dumas. 
Parisian aluminium bought in La Haag, 
Analyzed by SalvStat. Contained also a trace of sodium. 
Parisian aluminium bought in Bonn and analyzed by Dr. Kraut. 
Made at the works near Rouen, in 1858, from cryolite. Analyzed by 

Analyzed by Sauerwein. Contained also traces of lead and sodium. 
Analyzed by Morin. Average of several months' work. 
Analyzed by Kraut. Represents the best product of the French works 

sent to the London Exhibition in 1862. 
Analyzed by Mallet. The best metal v/hich could be bought in 1880. 

Purchased in Berlin by Mallet, and used by him as the material which 

he purified and used for determining the atomic weight of aluminium. 
Analyzed by Hampe. This was the purest metal which could be bought 

in 1876. No. 14 contained also per cent, of copper and 0.20 per 

cent, of lead. No. 15 contained 0.40 per cent, of copper and 0.20 per 

cent, of lead. 
Bought in Philadelphia as Frishmuth's aluminium, in 1885, and analyzed 

by the author. 


17. Specimen of the metal composing the tip of the Washington Monument, 

cast by Frishmuth. This analysis is reported by R. L. Packard m the 
Mineral Resources of the United States, 1883-4. 

18. The best ordinary grade of metal made by this process, analyzed by Hunt 

& Clapp, Pittsburgh. For average analyses, etc., see description of pro- 

19. The best grade made by this process, exhibited at the Paris Exposition, 

1889. Analyzed by CuUen. 

20. Analysis by Dr. Kraut of metal being made on a commercial scale. 

21. Analysis by Grabau of the purest metal yet obtained by his process. 

22. Made from chemically pure alumina, in order to determine what degree of 

purity was attainable by the electrolytic processes. 

In addition to the above, the following analyses made in the 
Pittsburgh Testing Laboratory show the quality of commercial 
aluminium which has been put on the market in recent years : 

Com- Graphi- 

Alumin- bined toidal 

Second Quality Metal. ium. Silicon. Silicon. Iron. Copper. Sodium. Lead. 

Pittsburgh Reduction Co 9S-00 1.50 1.35 2.00 0.07 0.04 0.03 

Society Anon, de TAluminium .. 95.00 0.90 i.oo 3.00 0.05 o.oi 0.04 

Alliance Aluminium Co 95-00 0.90 0.75 3.25 0.02 0.07 o.oi 

The Aluminium Co., Limited . . . 95.00 0.85 0.75 3.00 0.30 nil 

The Al. LA. G., Switzerland ... 95.00 1.75 1.15 2.00 0.07 nil 

" " " ... 94.15 1.02 0.54 2.80 1.49 — — 

First Quality Metal. 

Pittsburgh Reduction Co 97.00 1.55 1.25 0.13 0.03 0.02 0.02 

Soc. Anon, de I'Al 97.00 0.90 0.82 1.20 0.04 o.oi 0.03 

Alliance Al. Co., Ltd 97.00 0.95 0.53 1.45 0.01 0.05 o.oi 

The Al. Co., Limited 97.00 0.75 0.52 1.55 0.03 0.15 nil 

Pittsburgh Reduction Co 98.00 1.30 0.60 0.07 o.oi o.oi o.oi 

Soc. Anon, de I'Al 98.00 0.71 0.35 0.90 o.oi o.oi 0.02 

Alliance Al. Co., Ltd 98.00 0.90 0.29 0.75 o.oi 0.04 o.oi 

The Al. Co., Limited 98.00 0.90 0.23 0.80 0.02 0.05 nil 

Pittsburgh Reduction Co 98.52 0.42 0.72 0.05 0.06 nil 0.04 

" " " 99.00 0.80 0.15 0.03 O.OI nil O.OI 

Soc. Anon, de I'Al 99-00 0.35 0.13 0.50 nil o.oi o.oi 

Alhance Al. Co., Ltd 99.00 0.31 0.20 0.45 nil 0.03 o.oi 

The Al. Co., Limited 99-°° °-^l °-'S o-5S °-°2 o.oi nil 

Pittsburgh Reduction Co 99-20 041 0.34 0.05 nil nil nil 

Alliance Al. Co., Ltd 99- H o-23 0.17 0.46 nil nil nil 

Pittsburgh Reduction Co 99-34 °A° 0-21 0.05 nU nil nil 

The makers represented above are : 

The Pittsburgh Reduction Company. New Kensington, Pa. 


The Alliance Aluminium Company, Limited. Wallsend-on- 
Tyne, England. 

The Aluminium Company, Limited. Oldbury, near Birming- 
ham, England. 

The Societe Anonyme de I'Aluminium. Nanterre, France. 

The Aluminium Industrie Actien-Gesellschaft. Neuhausen, 

The second, third and fourth of these firms used the sodium 
process, the others the electrolyic methods. It is at once seen 
that for metal with the same proportion of aluminium, that 
made by the electrolytic processes contains less iron, but more 
silicon, than that made by the sodium processes ; there is no 
observable regularity in the proportion of combined and graph- 
itoidal silicon to the amount of iron. On an average, the graph- 
itoidal silicon is about one-third to two-thirds of the combined 

Rammelsberg (Kerl's Handbuch) investigated the question 
of the different states of silicon in aluminium. He found in 
aluminium reduced from cryolite by sodium in a porcelain cru- 
cible, that on treatment with hydrochloric acid, the combined 
silicon changes into silica, or part escapes as silicon hydride, 
while the graphitoidal silicon remains as a black residue. Two 
analyses gave him the following results : 

1.. 2. 

Silicon obtained as silica 9.55 1.85 

Free silicon 0.17 0.12 

Silicon escaping in SiH'' 0.74 0.58 

In aluminium containing 6.20 per cent, of total silicon, ana- 
lyzed in the Pittsburgh Reduction Co.'s laboratory, 3.85 per 
cent, was graphitoidal silicon. In another specimen, contain- 
ing 5 -30 per cent, of total silicon, 2.40 per cent, was graphi- 

An experiment, made at the Pittsburgh Testing Laboratory, 
to determine the influence of quick or slow cooling on the con- 
dition of the silicon, showed that the chilled metal contained, 
slightly more of the silicon in the graphitoidal state and less in 
the combined state than metal slowly cooled. 


This result is the opposite of the behavior of carbon in iron, 
and it is found, as might be expected, that no appreciable dif- 
ference in hardness can be detected between the surfaces of 
ingots cast in chills and those cast in hot moulds. 

M. Dumas found that aluminium usually contains gases, 
about which he makes the following statements:* On submit- 
ting aluminium in a vacuum to the action of a gradually increas- 
ing temperature up to the softening point of porcelain, and 
letting the mercury pump continue acting on the retort until it 
was completely exhausted, considerable quantities of gas were 
withdrawn. The liberation of the gas from the metal seems to 
take place suddenly towards a red-white heat. 200 grammes 
of aluminium, occupying 80 c.c, gave 89.5 c.c. of gas, meas- 
ured at 17° and 755 mm. pressure. The gas consisted of 1.5 
c.c. carbonic acid, and 88 c.c. hydrogen. Carbonic oxide, 
nitrogen and oxygen, were absent. 

The author has observed that molten aluminium will absorb 
large quantities of gas. On passing sulphuretted hydrogen 
into the melted metal for about twenty minutes some aluminium 
sulphide was formed, while the metal appeared to absorb the 
gas. On pouring, the metal ran very sluggishly, with a thick 
edge, but when just on the point of setting, gas was disengaged 
so actively that the crackling sound could be heard several feet 
away, and the thick metal became suddenly quite fluid and 
spread over the plate in a thin sheet. The gas disengaged 
seemed by its odor to contain a large proportion of sulphur- 
etted hydrogen, although free hydrogen may have been pres- 
ent in it. 

Le Verrier has established the presence of carbon in small 
amount in commercial aluminium.! Taking a large quantity of 
metal and treating it with a current of hydrochloric acid or 
hydriodic acid gas, entirely free from oxygen, there remains a 
grey residue. This, treated with dilute hydrochloric acid, gives 
a very light residue of amorphous carbon, of a maroon color, 

* Comptes Rendus, xc, 1027 (1880). 
t Comptes Rendus, July, 1894, p. 14. 


■which, on being separated out, can be entirely burnt to car- 
bonic acid gas in a current of oxygen. This carbon does not 
contain a trace of graphite ; it is entirely amorphous. It is 
easiest estimated by dissolving ten grammes of aluminium in a 
concentrated solution of caustic potash, washing the residue 
with water, drying, burning in a current of oxygen, and weigh- 
ing the carbonic acid formed. In three specimens tested there 
were found 0.104, O.108 and O.080 per cent, of carbon respect- 
ively. To determine the effect of carbon on the physical prop- 
erties of aluminium, he melted best quality metal, cast a 
specimen ingot, and then dissolved in it some crystallized 
aluminium carbide, AhCs (for method of making this see Chap- 
ter XIV) . A specimen of the carburized metal was then cast 
and tested. The results were as follows; the weights being per 
square millimetre of section : 

Cast Specimens. 

Elastic Limit. Breaking Load. Elongation. 

Pure aluminium — 1 1. 1 kilos. 9 per cent. 

Carburized aluminium — 6.5 " 5 " 

— 8.6 " 3 " 

Rolled Specimens. 

Elastic Limit. Breaking Load. Elongation. 

Pure aluminium, hard 23 kilos. 25.0 kilos. 2.0 per cent. 

" " annealed 8 " 14.0 " 23.0 " 

Carburized aluminium, hard 20 " 20.79 " 2.5 " 

" " annealed 7.7 " 13.80 " 26.5 " 

It would appear that the cast metal shows the greatest de- 
terioration by the presence of carbon. 

Le Verrier also proved the presence of traces of nitrogen in 
commercial aluminium by passing the hydrogen evolved during 
solution in dilute caustic potash into a Nessler solution, which 
then gives the ammonia reaction. To determine the exact 
effect of nitrogen, some aluminium was melted and a current of 
nitrogen passed into it. The metal became saturated with the 
gas, and showed the following diminution in strength : 


Cast Specimens. 

Elastic Limit, Breaking Load. Elongation. 

Pure aluminium 7.5 kilos. i I.I kilos. 9 per cent. 

Aluminium saturated with nitrogen 6.5 " g.6 " 6 " 

Le Verrier considers that nitrogen and carbon occur in alu- 
minium in the state of aluminium nitride and carbide re- 
spectively, dissolved in the excess of aluminium. The same 
investigator has discovfered by the microscope, in the residue 
left by dissolving in hydrochloric acid, very small, sharp crys- 
tals of boron carbide, the boron coming from the boric acid 
used in agglomerating the carbon electrodes. A little amor- 
phous alumina is also sometimes to be found. 


Deville : The color of aluminium is a beautiful white with a 
slight blue tint, especially when it has been strongly worked. 
Being put alongside silver, their color is sensibly the same. 
However, common silver, and especially that alloyed with 
copper, has a yellow tinge, making the aluminium look whiter 
by comparison. Tin is still yellower than silver, so that alu- 
minium possesses a color unlike any other useful metal. 

Mallet: Absolutely pure aluminium is perceptibly whiter 
than the commercial metal ; on a cut surface very nearly pure 
tin- white, without bluish tinge, as far as could be judged from 
the small pieces examined. 

The purest-looking aluminium examined by the author is 
that made by Grabau. On a fresh fracture it is absolutely 
■white, but on long exposure to the air it takes a faint, almost 
imperceptible bluish tint. On a cut surface it has the faintest 
suspicion of a yellow tint, not so decided as the yellowish color 
of pure tin. Ordinary commercial aluminium is bluish on a 
fresh fracture, the tint being deeper the greater the amount of 
impurities it contains. A specimen with 10 per cent, of silicon 
and 5 per cent, of iron was almost as blue as lead. It is my 
belief that a very small percentage of copper closes the grain 
-and whitens the fracture a little; I have also found that chilling 


suddenly from a high temperature has the same effect. Wheir 
ingots of aluminium are exposed a long time to damp air, the 
thin film of oxide forming on them gives a more decided bluish 
cast to the metal, since the coating is perfectly snow-white and 
hence, by contrast, heightens the bluish tint of the metallic 
background. Mourey recommended removing this discolora- 
tion by placing the articles first in dilute hydrofluoric acid, 
1,000 parts of water to 2 of acid, and afterwards dipping in 
nitric acid. The oxide would thus be dissolved and the original 
color restored. Pure aluminium possesses to the highest de- 
gree that property expressed best by the French term " eclat." 
It is rather difficult to see why the blue tint should be more 
prominent after the metal has been worked, yet I think two 
reasons will explain this phenomenon ; first, aluminium is not a 
hard metal, and on polishing or burnishing, particles of dirt or 
foreign substances are driven into the pores of the metal,, 
thereby altering its color slightly ; second, any metal looks 
whiter when its surface is slightly rough than when highly 
polished, in the latter case it being as much the reflected color 
of the general surroundings as the color of the metal itself 
which is seen. I have never seen any highly polished white 
metal which did not look bluish, especially when reflecting out- 
door light. I think this explains why opera-glasses, rings, 
jewelry, etc., generally look bluer than the bar or ingot-metal 
from which they are made. 

Aluminium takes a very beautiful mat, which keeps almost 
indefinitely in the air, the surface thus slightly roughened ap- 
pearing much whiter than the original polished surface. Alu- 
minium can be polished and burnished without much difficulty 
if attention is given to a few particulars which it is necessary to 
observe. (For methods of poHshing, etc., see Chapter XIII.) 

Extremely thin films of aluminium are fine violet-blue by 
transmitted light. 


A cast ingot of purest aluminium has a slightly fibrous 
structure, a section y^ inch thick bending twenty degrees or 


SO from a straight line when sharply bent before showing cracks 
at the outside of the turn. The fracture of such an ingot is 
uneven, rough, and very close, often showing a curious semi- 
fused appearance, as if it had been already exposed to heat and 
the sharpest points melted down. However, only the purest 
varieties show these peculiarities. Metal containing 96 to 97 
per cent, of aluminium begins to show a crystalline structure, 
breaks short, and with a tolerably level surface. Metal less 
than 95 per cent, pure shows large shining crystal surfaces on 
the fracture, the smaller crystals being on the outside of the 
ingot where it has been cooled most quickly, while in the cen- 
tre the crystalline surfaces may be as large as tV inch in 
diameter. A specimen containing only 85 per cent, of alumin- 
ium broke as short as a bar of antimonial lead, with a large 
granular, crystalline surface. 

Working the metal increases its fibrousness greatly, the sec- 
tion of a square rolled bar of good metal looking very much 
like that of a low-carbon steel. 


The purest aluminium is distinctly softer than the commer- 
cial ; estimated on the scale of hardness proposed by Mohs it 
would be written as about 2.5, that is, a little harder than can 
be scratched by the nail. It is not so soft as pure tin. The 
presence of impurities, however, rapidly increases the hardness. 
While 99 per cent, aluminium can be cut smoothly with the 
knife and shavings turned up almost as with pure tin, yet 95 
per cent, metal can hardly be cut at all ; the shavings break off 
short, and a fine grating is felt through the blade. 

Aluminium becomes surface hardened by cold working. 
Castings of aluminium drop-forged cold, in dies, are made 
sensibly harder and thus adapted for standing more wear. 
Cold drawing or rolling gives to it nearly the hardness of brass. 
In a drop test to investigate this quality made in the Pittsburgh 
Testing Laboratory, the following comparative results were 
reached : 



Aluminium, rolled 62 

Aluminium, cast 47 

Brass, rolled 86 

Brass, cast 71 

Copper 100 

Experience in testing various specimens of commercial alu- 
minium with the knife will, I am sure, enable a person to be- 
come quite skillful in determining the purity and in separating 
different grades from each other. Taking this test in connec- 
tion with the breaking and surface of fracture, it appears to me 
that these indications are as significant and can be made of as 
much use as the corresponding tests for iron, steel, and other 
metals. Mr. Joseph Richards, the author's father, having had 
many years' experience in testing lead, tin, zinc, and similar 
metals, in which the knife-blade has been put to good service, 
has been able with very little practice to arrange a number of 
specimens of aluminium correctly according to their purity sim- 
ply by noting carefully the way they cut and the color of the 
cut surface. These tests will in the future, I am sure, be of 
great use to those handling aluminium on a large scale, es-- 
pecially in the works where it is produced. 

Specific Gravity. 

Mallet: The specific gravity of absolutely pure aluminium- 
was carefully determined at 4° C, and the mean of three 
closely agreeing observations gave 2.583. 

Commercial aluminium is almost always heavier than this, 
but the increase is not in direct proportion to the amount of 
impurities present. There are two reasons why this last state- 
ment is correct; first, we cannot say what expansion or con- 
traction may take place in forming the alloy; second, while- 
most of the impurities which occur are much heavier than alu- 
minium, yet silicon, the most frequent of all, has a specific 
gravity of only 2.34 (Deville's determination), and therefore 
acts in the opposite direction to the other impurities, though. 


not to SO great an extent. The following analyses and specific 
gravities may give some information on this point: — 


Specific Gravity. 










2.61 (2.64) 





2.61 (2.69) 





2.59 (2.74) 






It is seen in each case that the calculated specific gravity is. 
much less than the observed, which would show contraction in 
volume by alloying. Indeed, this is a prominent characteristic 
of aluminium alloys, aluminium often taking up several per 
cent, of its weight of another metal without its volume being 
increased, the particles of the other metal seeming to pass be- 
tween those of the aluminium ; thus probably accounting for 
the extraordinary strength and closeness of many of the alumin- 
ium alloys. This subject is treated more at length in the chap- 
ter on alloys. We can see the large contraction taking place 
by inspecting the numbers in parentheses under the heading 
"Calculated." These are computed on the supposition that 
the volume of the impure aluminium is equal to that of the 
pure aluminium entering into it. As these numbers are also- 
less than the observed specific gravities, the extraordinary fact 
is shown that aluminium can absorb several per cent, of iron 
and silicon and yet will decrease in volume in doing so. 

The remarks thus far made are based on the gravity of cast 
metal. Aluminium increases in density by being worked : De- 
ville states that metal with a specific gravity of 2.56 had this 
increased to 2.67 by rolling, which, he says, may explain the 
diflerences existing in its properties after being annealed or 
worked. He remarked further that heating this rolled metal to 
100°, and cooling quickly, changed its specific gravity very little, 
lowering it to 2.65. I have observed that on heating a piece of 
aluminium almost to its fusing point and suddenly chiUing it in 
water, its specific gravity was lowered from 2.73 to 2.69. 


Experiments made by Prof. J. W. Langley with aluminium 
98.52 per cent, pure, the rest being mostly silicon, gave its 
specific gravity as follows : 

Sample sawn from centre of a cast ingot 2.587 

Same, annealed 2.692 

Rolled into sheet 0.0625 inch thick 2.710 

Stamped into medals 2.725 

The low specific gravity of aluminium, when compared with 
that of the other metals, is (in the words of a recent lecturer) 
" the physical property on which our hopes of the future use- 
fulness of aluminium chiefly rest." The following table will 
facilitate this comparison : 

Specific Gravity. 

Alumin- Founds in a Kilos in .1 

Water = I. ium =: i. cubic foot. cubic meter. 

Platinum 21.5 8.3 1344 21,500 

Gold 19.3 7.4 1206 19,300 

Lead 11.4 4.6 712 11,400 

Silver 10.5 4.0 656 10,500 

Copper 8.9 3.5 557 8,900 

Iron and steel 7.8 2.8 487 7,800 

Tin 7.3 2.7 456 7,300 

Zinc 7.1 2.7 444 7,100 

Aluminium 2.6 i.o 163 2,600 

In comparing the price of aluminium with that of a metal 
which it might replace, for purposes where the bulk of the arti- 
cle is fixed, it is of first importance to take this low specific 
gravity into account. When we come to consider closely, we 
find that almost every article of use or ornament is wanted of a 
fixed size, and therefore the real basis of comparison of prices is 
that of a fixed volume of the metal rather than a fixed weight. 
To facilitate this comparison, I give below the actual market 
prices of one pound weight of the common metals, with the 
prices of equal volumes: 


Value of Relative cost Value of Relative cost 

I pound, of equal weights, i cubic foot, of equal volumes. 

Silver jSio.oo 2850 ;J656o.oo 1 1430 

Nickel 0.50 140 279.00 468 

Aluminium 0.35 100 S7-S0 10° 

German silver 0.30 85 167,00 290 

Tin 0.15 43 68.50 120 

Bronze 0.15 43 79.00 137 

Brass 0.12 34 61.75 107 

Copper 28 55.75 97 

Lead 0.04 11 28.50 50 

Zinc 0.04 II 17-75 3' 

Steel 0.02 6 9.75 17 

Cast Iron o.oi 3 5.00 8 

It is therefore evident, that for almost all uses to which 
aluminium is applicable, there are only five of the common 
metals which are cheaper, viz., copper, lead, zinc, steel and cast- 
iron. When it is further remembered that the cost of the metal 
alone usually forms only a small part of the value of the finished 
article, it can be further asserted that there should be only a 
trifling diflference between the cost of an article made in alu- 
minium and one of tin, bronze, brass or copper. The sole 
reason why aluminium articles are not sold more cheaply at 
present, lies in the fact that they are not made and sold in such 
large quantities ; the increase of the aluminium industry will 
therefore ensure lower prices for aluminium goods. 


Deville : Aluminium melts at a temperature higher than that 
of zinc, lower than that of silver, but approaching nearer to that 
of zinc than silver. It is, therefore, quite a fusible metal. 

Mallet : It seems that pure aluminium is a Httle less fusible 
than the commercial metal. 

Pictet determined the melting point to be 600°, Heeren about 
700°, while Van der Weyde placed it as high as 850°. On a 
sample containing 0.5 per cent, of iron, Carnelly determined 
the melting point at 700° ; with 5 per cent, of iron it did not 
fuse completely until 730°. The discordance between these 
various experimenters lies chiefly in the errors of the pyro- 


meters used. Since the perfection of the electric pyrometer by 
Le ChateHer, the melting point of the purest aluminium has 
been repeatedly observed as very close to 625°. This figure 
is probably correct within 5°. 

A very small admixture of iron and silicon lowers the melt- 
ing point. Le Chatelier found with a metal containing 1.50 
per cent, silicon and 0.75 per cent, iron, the melting point 
619°- A larger quantity of either of these elements quickly 
operates in the other direction, and prevents fluid fusion until 
a temperature above 625° is reached. 

Spring has determined that aluminium powder flows to a 
solid mass at a pressure of 38 tons per square inch, at ordinary 


Deville : Aluminium is absolutely fixed, and loses no part of 
its weight when it is violently heated in a forge fire in a carbon 

This statement was made in 1859, and can still be accepted 
as true as far as ordinary furnace temperatures are concerned. 
But, with the use of the electric furnace, temperatures have 
been attained at which aluminium does sensibly volatilize. In 
Cowles Bros, electric furnace it is stated that the aluminium is 
almost all produced as vapor and as such is absorbed by the 
copper or iron present ; when these are not present it is found 
condensed in the cooler upper part of the furnace. A similar 
experience has been met in other electric furnace processes, so 
that the volatilization of aluminium at these extreme tempera- 
tures may be accepted as a fact. 

Aluminium vapor has been recently produced in an electric 
furnace in large enough amounts to determine that it is of a 
green-blue color. 

The purest aluminium has no perceptible odor, but metal 
containing i per cent, or more of silicon exhales the odor of 
silicon hydride, exactly the same odor as proceeds from cast 


iron. Second quality commercial aluminium gives this smell 
quite strongly, but in the present first quality metal it is difficult 
to detect it except by smelling immediately after a brisk 

The same remark may be made regarding taste as of the 
odor. The purest metal has no taste ; the impure tastes like 
iron, but not so strongly. 


Deville, as also Poggendorff and Reiss, state that aluminium" 
is very feebly magnetic. It must be remembered, however, that 
they tested metal made by the sodium process, which almost 
invariably contained iron. 

Prof. Frank Very, of the Allegheny Observatory, Allegheny, 
Pa., finds aluminium with only a trace of iron to be absolutely 
non-magnetic. With 0.05 per cent, of iron, none could be ob- 
served. An ingot containing 1.5 per cent, iron showed faint 
polarity, while with 2 per cent, the polarity was very decidedly 

Since the present first quality commercial aluminium contains 
less than 0.2 per cent, of iron, and frequently only 0.05 per 
cent., we may observe that this is less than is usually found in 
commercial brass, bronze or german-silver, and for this reason, 
as well as its other valuable properties, aluminium answers very 
well for compass boxes or cases, or parts of electrical instru- 
ments where non-magnetic properties are desired. 

Deville: "A very curious property, which aluminium shows 
the more the purer it is, is its excessive sonorousness, so that 
a bar of it suspended by a fine wire and struck sounds like a 
crystal bell. M. Lissajous, who with me observed this prop- 
erty, has taken advantage of it to construct tuning forks of 
aluminium, which vibrate very well. I also tried to cast a bell, 
which has been sent to the Royal Institution at London at the 


request of my friend Rev. J. Barlow, vice-president and secre- 
tary of the institution. This bell, cast on a model not well 
adapted to the qualities of the metal, gives a sharp sound of 
considerable intensity, but which is not prolonged, as if the 
clapper or support hindered the sound, which, thus hindered, 
becomes far from agreeable. The sound produced by the in- 
gots is, on the contrary, very pure and prolonged. In the ex- 
periments made in Mr. Faraday's laboratory, this celebrated 
physicist has remarked that the sound produced by an ingot of 
aluminium is not simple. One can distinguish, by turning the 
vibrating ingot, two sounds very near together and succeeding 
each other rapidly, according as one or the other face of the 
ingot faces the observer." 

The bell referred to above was 20 kilos in weight and 50 
centimetres in diameter, but as Deville admits, its sound was 
not pleasing, and a contemporary writer, evidently not very 
enthusiastic in sounding the praises of aluminium, said that 
" while the bell was highly sonorous, yet it gave a sound like 
a cracked pot." 

I have not heard that any large bell has since been cast, but 
it is certain that the metal in bars has a highly musical ring. 
Faraday's observation has also been verified, for a recent lec- 
turer suspended by one end a bar 6 feet long 31^ inches wide, 
and I y^ inches thick, and on striking it a prolonged vibration 
ensued, two notes being recognized, A sharp and D sharp, the 
latter more subdued. 

Within the last few years aluminium has been extensively 
employed for sounding boards in public hails, the sounding 
boards of pianos, and even for the construction of horns and 
violins. The sounding board in the Lyceum Hall, New York 
city, is an example. The particular advantage is, its not being 
affected by a moist atmosphere. Dr. Alfred Springer, of Cin- 
cinnati, in speaking of its use for sounding boards says,* " It 
.differs from all other metals in the absence of the comparatively 

* Trans. A. A. Adv. Science, 1891, vol. 40, p. 182. 


continuous and uniform higher partial tones, which give in 
other metals the tone-color called " metallic," and further, that 
it also possesses an elasticity capable of sympathetic vibration 
uniformly through a wide range of tone-pitch, which renders it 
in this respect superior to wood." 

Dufour has measured the velocity of sound in aluminium at 
4950 metres per second, almost the same figure as for steel. 

Crystalline Form. 

Deville : " Aluminium often presents a crystalline appearance 
when it has been cooled slowly. When it is not pure, the little 
crystals which form are needles, and cross each other in all 
directions. When it is almost pure it still crystallizes by fusion, 
but with difficulty, and one may observe on the surface of the 
ingots hexagons which appear regularly" parallel along lines 
which centre in the middle of the polygon. It is an error to 
conclude from this observation that the metal crystallizes in the 
rhombohedral system. It is evident that a crystal of the regu- 
lar system may present a hexagonal section; while on the 
other hand, in preparing aluminium by the battery at a low 
temperature, I have observed complete octahedrons which were 
impossible of measurement, it is true, but their angles appeared 

By slowly cooling a large body of melted aluminium and 
pouring out the fluid interior, distinct octahedrons may be 


Cast aluminium is not very elastic; it may be likened in this 
respect to cast silver. When worked, however, it becomes 
more rigid and elastic. 

Mallet remarked that absolutely pure aluminium seemed to 
be less hardened by hammering than ordinary commercial 

Cast aluminium stiffens up very quickly in rolling or drawing, 
and the strains set up in the metal can be removed by an 


annealing, which consists in heating the metal to an incipient 
red heat and then either allowing to cool slowly, or leaving a 
few minutes at that heat and then cooling quickly by plunging 
into water. Metal thus treated becomes very soft and pliable, 
hardly any more elastic than lead, and will stand considerable 
working before it again becomes stiff and elastic. The presence 
of 2 to 5 per cent, of copper, nickel, silver, German-silver or 
titanium increases very greatly the elasticity of aluminium 
without increasing noticeably its specific gravity or impairing 
its malleability. 

Tensile and Compressive Strength. 

W. H. Barlow* was the first to carefully test the tensile 
strength of aluminium. He used a rolled bar one-quarter 
inch square. With -a test piece two inches long, the tensile 
strength was found to be 26,800 pounds per square inch, with 
elongation before breaking of 2.5 per cent. 

Kamarschf tested aluminium wire, and obtained an average 
of 12.5 kilos per square millimetre, equal to nearly 18,000 
pounds per square inch. This wire must have been annealed, 
while the bar tested by Barlow was worked stiff before testing. 

Mr. Spilsbury, of the Trenton Iron Co., obtained with alu- 
minium wire of one-sixteenth inch diameter a tensile strength 
of 62,300 pounds per square inch, the wire having been drawn 
with one annealing from a wire-rod one-sixth inch in diameter. 
The same wire made 47 bends at a right angle before breaking. 

A most valuable series of mechanical tests of aluminium 
have been made by the Pittsburgh Reduction Company, and 
are copied as follows| : 

* Rpt. British Assoc. Adv. Science, 1882, p. 668. 

fDingler, Vol. 172, p. 55. 

J Transactions Am. Inst. Mining Engineers. Vol. XVIII., p. 528 (1890). 

Tensile Tests. 


No. I quality commercial metal, averaging 
over 98 per cent. pure. 

Ordinary castings of over \ inch sectional 

area, as cast .... 

Same, annealed 

Same castings, hardened by drop forging cold 

Rolled bars, left hard 

Same, annealed 

Hammered bars, left hard 

Same, annealed 

"Wire i to J inch diameter, as drawn 

Same, annealed 

Wire y\ to -^^ inch diameter, as drawn 

Wire ^ to ^ J ^ inch diameter, as drawn 

Same, annealed 

Plates to \ inch thick, as rolled 

Same, annealed 

Sheets \ to -^-^ inch thick, as rolled 

Same, annealed 

Sheets ^V '° ttss '^^'^ 'C!\\<^, as rolled 

Same, annealed 

•a 3 § 












»H 4) u 

■a p-.s 

rt ^ rt 












•a . 

& O 

a a 

4) j-j • 

"S "> So 





















Compressive Tests. 

Same metal is in above tensile 
tests, about 98 per cent. pure. 

Ordinary casting, as cast 

Same, annealed 

Same, hardened by drop forging. 

Rolled bars, left hard 

Same, annealed 

Hammered bars, left hard 

Same, annealed 

Metal containing 4.5 per cent, 
silicon and 1.5 per cent, iron, 
casting, unannealed 

Metal containing 1.14 per cent, 
silicon and 0.05 per cent, iron, 
castings, unannealed 

Same, larger test-piece 


DA . 



•= 0. c 

"»-S a 

e »■" 


0, a <u 

n po 

^^ .« (fl 























'u a 

Height X diam. 

I in. X 0.5 in. 
I in. X 0.5 in. 
I in. X 0.5 in. 

1.5 in. X 0.75 in. 
1.5 in. X 0.75 in. 

m. XI.5 m. 
in. X 1.5 in. 

I m. X0.5 m. 

I in.xo.5 in. 
3 in. X 1.5 in. 

Height X diam. 

0.985 X 0.503 
0.975 * 0.506 
0.965 X 0.508 


2.91 X 1.520 
2.91 X 1.519 

0.965 X 0.508 

0.975 X 0-507 

2.92 X 1.519 

Transverse Tests. 

Metal containing 98.52 per cent, aluminium, 
1. 14 per cent, silicon, and 0.05 per cent. iron. 

Test-piece one inch square, supported on knife 
edges 24 inches apart, loaded in the centre . . . 

Same metal, test-piece a 1.5 inch diameter rolled 
bar, on knife-edges 12 inches apart, loaded in 
the centre 






set, inches. 















































From the various tests above enumerated, Messrs. Hunt,. 
Langley and Hall draw the following general conclusions : 

The average strength of commercial aluminium of the follow 
ing average composition : 

Aluminium 97. to 99. per cent. 

Graphitic silicon o.i to i.o " 

Combined " 0.9 to 2.8 " 

Iron 0.04 to 0.2 " 

is as follows : 

Elastic limit in tension. 

i- Castings. 

Sheet . . . 

I Bars 

Pounds per square inch. 



16,000-30,000 (according to fineness)' 


Ultimate strength in ten- 

30,000-65,000 (according to fineness) 

Percentage of reduction 

Castings . 


iction j 
of area, in tension .... ] Wire 
L Bars. 

15 per cent.. 

35 " 
60 " 
40 " 


Elastic limit in compression in cylinders with length twice the diameter 3,50o 

Ultimate strength ■ 12,000 

Modulus of elasticity, castings 1 1,000,000 

" " cold-drawn wire 19,000,000 

" " sheets and bars 13,000,000 

Professor Le Chatelier * has shown that the tensile strength 
of annealed aluminium wire varies with the temperature as 
follows : 

Kilos per sq. millimetre. Pounds per sq. inch. 

At 0° C. 1 8 26,000 

150° 13 18,500 

250° 7 1 0,000 

300° 5 7^000 

40o<^ 2 3,000 

* Lejeal's Aluminium. 


Minet found the following properties for aluminium strongly 
worked and annealed : 

Tensile strength. Elongation. 
Kilos per sq. mm. Pounds per sq. in. Per cent. 

After working 23.5 33iOOO 3 

After being annealed I hour at 400° 15.1 21,500 17 

He concludes from these results that, by careful annealing at 
proportionate temperatures and for proportionate times, any 
given strength between these limits may be obtained, or any 
■desired elasticity from 3 to 17 per cent. 

The results obtained for the tensile strength of aluminium 
show, in general, that it is greatly influenced by the amount of 
work done on the metal ; that while the pure, unworked metal 
is by no means rigid, but elongates 10 to 20 per cent, before 
breaking, as it is worked it becomes stififer and more elastic, its 
strength may be nearly doubled, but it becomes at the same 
time more brittle, as is shown by its elongation falling to 3 or 5 
per cent. After working, increased elongation can be given to 
it by annealing, but at the expense of ultimate strength. The 
annealed material, however, would be a safer rhaterial to use in 
any place where sudden shocks had to be sustained, if worked 
within its elastic limit. 

Taking the tensile strength of aluminium in relation to its 
weight, it shows a very high mechanical value ; in fact, the only 
material which can compare with it in tensile strength, weight 
for weight, is the finest steel or aluminium bronze. A com- 
parison on this basis is frequently made, in order to determine 
the material which gives the greatest strength for a given 
weight, or, rather, a given strength with the least weight. For 
instance, we might make the comparison by calculating the 
weight of rods of the different metals which would support a 
certain maximum strain before breaking, and from these num- 
bers derive a table of the relative weights of the metals to give 
a certain strength : 


Tensile strength, Relative strength Relative strength 

Rolled Bars or Large Wire. pounds per sq. in. for same section, for same weight. 

Aluminium 30,000 100 100 

Cast steel 100,000 333 135 

Soft steel 75,000 250 87 

Wrought iron S5,oco 183 63 

Aluminium bronze ....... 100,000 333 118 

" brass 60,000 200 70 

Gun bronze 35,000 117 37 

Red brass 45,000 150 47 

Copper 33,000 1 10 33 

It results from this comparison that only steel of the highest 
quality and the best aluminium bronze will give a greater 
strength for a given weight than aluminium. Steel, however, 
is subject to rust, whereas aluminium is not ; aluminium bronze 
itself is a metal requiring considerable skill in working it to 
always obtain the highest results ; these two are the only com- 
petitors which aluminium has for constructions where great 
strength combined with lightness is the first requirement, and 
expense is a secondary consideration. 


Aluminium stands high in the list of malleable metals. It 
can be forged or rolled as easily as gold or silver, and can be 
beaten out into leaf. 

Mallet found that absolutely pure aluminium is decidedly 
more malleable than the commercial metal, and is less hardened 
by working. Experience in rolling it on a large scale has since 
shown that the more impurity in the metal the harder it is to 
work, and the quicker it will become hard and need annealing. 
Some impurities, however, are more detrimental than others. 
One per cent, of iron renders the metal stronger and harder to 
roll, but more than this amount renders it less and less malle- 
able, until with 6 per. cent, it is crystalline and brittle. On the 
other hand, silicon may occur alone up to nearly 10 per cent, 
without rendering the metal unfit for rolling. 

Aluminium is most malleable between 100° and 150° C, and 
can be worked some time at that temperature before becoming 


hard. If rolled or worked cold, it needs more frequent anneal- 
ing. While annealed aluminium is one of the softest and most 
pliable metals, yet by cold working it can be given the stiffness 
and temper of hard brass. The best commercial metal is 
regularly rolled into sheets one-thousandth of an inch in thick- 
ness, which sheets are sold for beating purposes, and it has 
been rolled to one-half that thickness. 

Metal for beating into leaf must be of the finest quality, i. e., 
over 99 per cent. pure. Rousseau, of Paris, was the first to 
make aluminium leaf, and it was afterwards made for several 
years by C. Falk & Co., of Vienna. At present, Mr. Kemp, of 
New York, manufactures it on a large scale. The beating is a 
little harder than is required for gold and silver, but by starting 
with the purest aluminium equally thin leaf can be obtained. 
A specimen of commercial aluminium leaf examined by the 
writer had a thickness of 0.000638 millimetre, or one forty- 
thousandth of an inch, which compares favorably with ordinary- 
silver leaf. This leaf was opaque, but the thinnest attainable 
leaf is a fine blue by transmitted light. No other useful metals 
except gold and silver will show such malleability. This leaf is 
now in regular use with gilders and decorators, and has en- 
tirely superseded silver leaf because of its cheapness and non- 
tarnishing qualities. The leaf is also rubbed into powder, and 
in that state used as a metallic dust in printing and decorating. 


Aluminium can be drawn into very fine wire. Even as early 
as 1855, M. Vangeois made very fine wires with metal far from 
being pure, and used them for making aluminium passementerie. 
The metal quickly hardens at the drawing plate, and must be 
frequently annealed. The larger sizes of wire can be drawn 
warm, and then require less frequent annealing. The heated 
gases from the chimney of an Argand burner will anneal the 
very fine wires if they are simply passed across it. Wire as 
fine as o.i millimetre (one two-hundred-and-fiftieth of an inch) 
can be made without much trouble, while wire one-seventieth 


of an inch in diameter is regularly kept in stock by the alu- 
minium dealers. 

Gold, silver, platinum, iron and copper are probably the only 
metals more ductile than aluminium. 

Expansion by Heat. 

Fizeau determined that aluminium expands 0.00222 of its 
length in passing from the freezing point to the boiling point, 
or 0.0000222 of its length for a rise of 1° Centigrade, equal to 
O.OOOOI129 per 1° Fahrenheit. 

Prof. J. W. Langley examined metal 98.5 per cent, pure, the 
rest being mostly silicon, and obtained for the expansion for 
1° C. 0.0000206, which would be o.ooooi 15 for 1° F. 

A bar of aluminium i foot long at 0° C. would, according to 
Fizeau, be 1.00222 feet long at 100°, while a cubic foot of 
aluminium at 0° would occupy 1.00666 cubic feet at 100° C, 
since the rate of cubical increase is three times that of linear 
expansion. This rate of expansion is close to that of tin, 
among the common metals. 

Specific Heat.* 

The best experimenters have given various values for the 
specific heat of aluminium, but it is very evident to the writer 
that the differences are mainly due to differences in the purity 
of the metal used. Professor Mallet used absolutely pure alu- 
minium (purified for atomic weight determination), and the 
writer used metal which analyzed 99.93 per cent, pure, and con- 
tained but a trace of iron, while the other experimenters used 
■ordinary commercial metal made by the sodium process and 
■containing considerable quantities of iron. 

Regnault, in 1855, used metal containing only 88.35 percent, 
of aluminium. His figure of 0.2056, between 25° and 97°, is, 
therefore, only approximate. 

* For further details, see papers on this subject by the author in Journal of the 
franklin Institute, April, 1892, and July, 1893. 


In 1856, Regnault obtained metal 97 per cent, pure, with 
which he obtained O.2122 (14° to 97°). 

Kopp, in 1863, used Parisian aluminium containing two per 
cent, of iron, and obtained a value varying between O.1970 and 
0.2070 (20° to 52°). His results are unreliable. 

In 1882, Mallet found the mean specific heat from 0° to 100°, 
using absolutely pure aluminium, 0.2253. 

Tomlinson * used commercial aluminium in wire, and gives 
the following formula for the mean specific heat, from any 
given temperature to 0° C. : 

Sm = 0.2070 4" O.OOOI 151. 

This would give from 0° to 100°, 0.2185. 
Naccari f used commercial metal, and gives the following for- 
mula, which expresses his results up to 300° C. : 

Sm = 0.2116 + 0.0000475 *• 

This would give from 0° to 100°, 0.2 164. 

Le VerrierJ stated that he found, using best commercial 
metal, the specific heat between 0° and 300° to be invariable, 
and = 0.22 ; between 300° and 530° also constant, and = 0.30; 
between 530° and 560° an absorption of heat, rendered latent, 
of about 10 calories; between 540° and 600° the specific heat 
again constant, and = 0.46. For the total quantities of heat 
in the metal, at various temperatures, he gives : 

o'' to 300° Q = 0.22 t. 

300° to 530° Q = 65 + 0.30 (t— 300). 

540° to 600° Q = 139 -\- 0.46 (t — 530). 

About 600° Q = 170. 

Before fusion at 620° Q = 200. 

These determinations would give the mean specific heat 
from 0° to 100°, 0.2200. No other investigator, however, has 
noticed the sudden jumps noted by Le Verrier, which, if true, 

*Proc. Royal Society, 1885, p. 494. 

t Trans. Accademia di Torino, Dec, 1887. 

I Comptes Rendus, 1892, Vol. 114, p. 907. 


are indeed very singular. Experiments by Pionchon and the 
writer, about to be described, gave no indications of such criti- 
cal points. 

Pionchon* states that he used very pure aluminium, and ob- 
tained the following formulae : 

291-86 t 
0° to 580° Q = 0.393 t- giy.8 4- 1 

625° to 800° Q = 0.308 t— 46.9. 


This would give the specific heat at 0°, = 0.2010; the 
mean specific heat 0° to 100° = 0.2 130; heat in solid alu- 
minium at its melting-point, 160.5 calories; heat in molten 
aluminium at the setting-point, 239.4 calories ; specific heat of 
molten aluminium constant, and = 0.308. The latent heat of 
fusion would be 239.4 — 160.5 = 7^-9 calories. 

The writer, using aluminium which analyzed 99.93 per cent, 
pure, with only a trace of iron and 0.07 per cent, silicon, ob- 
tained the following formulae for solid aluminium up to 600° : 

Mean specific heat to o°- Sm = 0.2220 -|- 0.00005 t. 

True specific heat at t°- S = 0.2220 -\- o.ocoi t. 

Total heat to o°- Q = 0.2220 1 -|- 0.00005 t'. 

These formulae give for the specific heat at 0° =0.2220; 
the mean specific heat 0° to 100° = 0.2270; the heat in solid 
aluminium at the melting-point, 158.3 calories; specific heat at 
the melting-point, 0.285. 

Regarding these figures, it will be observed that the mean 
value 0° to 100° is less than i per cent, different from Mallet's 
result, and therefore these values may be accepted as the true 
value of the mean specific heat between those temperatures. 
Regarding the value for the heat contained at the melting-point, 
it is only a fraction over i per cent, removed from that given 
by Pionchon's formula. It must not be forgotten, however, 
that aluminium begins to soften at about 600°, and some of its 
latent heat of fusion is absorbed before the true melting-point, 

*Comptes Rendus, 1892, Vol. 115, p. 163. 


625°, is reached, as observed by both Pionchon and Le Verrier. 
The numbers given by the formulae are the theoretical amounts 
of heat which the metal would contain at 625°, provided that 
none of the latent heat of fusion had been previously absorbed. 
To determine the amount of heat in the molten metal at its 
setting point the writer made several determinations by pour- 
ing the metal into water, just as it was setting. The result 
showed that the purity of the metal had a great effect on the 
amount of heat thus given up. 

Aluminium 96.9 per cent, pure gave up 229.0 calories. 

99-9 " " " 254-0 " 

99-93 " " " 258.3 " 

It is seen that a small amount of impurity makes a large 
difference, which is really caused by the impurity lowering the 
melting point of the metal. Since the metal used by Pionchon 
was not over 99.7 per cent, pure, we have a partial explanation 
of why he obtained only 239.4 calories in the molten metal. 
The figure deduced by the writer for the latent heat of fusion 
of aluminium is therefore 258.3 — 158.3=100 calories. 

The value of the specific heat of aluminium determined by 
Mallet and the writer agrees best with Dulong and Petit's law, 
as is seen by the following comparison of the actual and atomic 
specific heats for the interval 0° to 100° : 

Experimenter. Specific Heat x Atomic Weight= Atomic Heat. 
Regnault 0.2122 x 27.0 = 5.73 

Kopp 0.2020 

Tomlinson 0.2185 

Naccari 0.2164 

Le Verrier 0.2200 

Pionchon 0.21 30 

Mallet 0.2253 

Richards . 0.2270 



Since for all the common metals the product of the atomic 
weight by the specific heat for this range of temperature is in- 
variably above 6.0, and averages 6.2, the conclusion is plain 
that the last two figures given above must be the correct ones. 


The high value of the specific heat of aluminium compared 
with other metals, and particularly its large latent heat of 
fusion, cause the metal to melt very slowly even in a very hot 
fire. As seen above, over 258 calories have to be absorbed 
in order to raise it to the melting point and melt it. It com- 
pares with other metals as follows : 

Aluminium 25^-3 calories. 

Cast iron 250.0 " 

Copper 162.0 " 

Platinum 102.4 " 

Silver 85.0 

Gold 58.0 " 

We can therefore deduce the interesting fact, which is amply 
confirmed by observation, that having a furnace hot enough to 
melt cast iron, we would find that a given weight of gold, silver 
or copper would melt before the same weight of aluminium, 
and even the cast-iron would not be far from melting by the 
time the aluminium fused. 

Electric Conductivity. 

Aluminium is a very good conductor of electricity. Metal 
used for electrical purposes should be of the best quality, and as 
free as possible from intermingled slag or blow-holes, which 
impair greatly its conductivity. 

Taking the electrical conductivity of absolutely pure silver or 
copper as 100, we cite the following determinations : 

Silver = 100. Copper = 100. 

Mattheissen. Commercial aluminium 33-76 

Benoit. Annealed wire 49-7° 

Watts 5610 

*Lorenz 49.1 at 0°. 

■< 50.1 at 100°. 

'■ Unnealed wire 54.8 at 14°. 

" 53-7 at 76°. 

Annealed " S7.4 at 14°. 

" " SS-9 at 76°. 

Weiler. Best commercial metal S4-20 54.3 

tC. K. McGee. Metal used 
98.52 per cent. pure. 

* Wied. Annalen, | 2], xiii, 422, 582 (1881). 
t Trans. Am. Inst. Mining Engineers, xviii, 528 (1890). 


The later determinations of McGee and Weiler agree very 
satisfactorily. Taking 54.8 as the relative resistance at ordinary 
temperatures, a wire of aluminium o.i inch in diameter and lOO 
feet long would have a resistance of 0.20235 ohms. 

Professor Dewar has recently investigated the electrical re- 
sistance of the metals at very low temperatures. He finds with 
aluminium, as with other metals, that the resistance varies 
directly as the absolute temperature. The following are his 
results : 

Relative Electrical Resistances. 

Temperature, Pure Silver. Pure Copper. Aluminium. 

C° .... .... 99 per cent. 97.5 per cent. 

ig.i 1588 1682 2772 2869 

i.o 2583 2683 

—82 1704 1738 

—197 .... .... 560 506 

—219 .... .... .... 325 

Thermal Conductivity. 

Faraday stated that he had found by a very simple experi- 
ment that aluminium conducted heat better than either silver 
or copper. 

L. Lorenz of Copenhagen, in the Jahresbericht der Chemie, 
1881, p. 94, stated that he had found it inferior to both of these 
metals, and later investigations have confirmed this. 

Professor Carhart, of the University of Michigan, finds that 
annealed aluminium is a slightly better heat conductor than 
the unannealed. 

The various results are as follows : 

Silver = 100, Copper = loo. 

L. Lorenz 47-72 at 0°. 

L. Lorenz ; . 50.00 at 100°. 

Calvert and Johnson 66.5 

Prof. Carhart. Annealed Metal 38.87 52.8 

Prof. Carhart. Unannealed Metal 37-96 51.6 

Gold is the only other metal which conducts heat better than 



We would here repeat the remark made with regard to the 
physical properties, that the properties to be recorded are those 
of the purest metal unless specifically stated otherwise. How- 
ever, the high grade of commercial metal differs very little in 
most of its chemical properties from the absolutely pure, so 
that not many reservations are necessary in applying the fol- 
lowing properties to good, commercial metal: 

Atomic Weight. 

The most accurate determination which we have is that of 
Professor Mallet, who found 27.02. For all practical purposes, 
and even in analyses, it may be conveniently considered as 27. 
The chemical equivalent weight is one-third of this, or 9. 

Action of Air. 

Deville: "Air, wet or dry, has absolutely no action on alu- 
minium. No observation which has come to my knowledge 
is contrary to this assertion, which may easily be proved by 
any one. I have known of beams of balances, weights, plaques, 
polished leaf, reflectors, etc., of the metal exposed for months 
to moist air and sulphur vapors and showing no traoe of alter- 
ation. We know that aluminium may be melted in the air with 
impunity, therefore air and also oxygen cannot sensibly afifect 
it. It resisted oxidation in the air at the highest heat I could 
produce in a cupel furnace, a heat much higher than that re- 
quired for the assay of gold. This experiment is interesting, 
especially when the metallic button is covered with a layer of 
oxide which tarnishes it, the expansion of the metal causing 



small branches to shoot from its surface, which are very bril- 
liant and do not lose their lustre in spite of the oxidizing at- 
mosphere. M. Wohler has also observed this property on try- 
ing to melt the metal with a blowpipe. M. Peligot has profited 
by it to cupel aluminium. I have seen buttons of impure 
metal cupelled with lead and become very malleable. 

"With pure aluminium the resistance of the metal to direct 
oxidation is so considerable that at the melting point of plati- 
num it is hardly appreciably touched, and does not lose its 
lustre. It is well known that the more oxidizable metals take 
this property away from it. But silicon itself, which is much 
less oxidizable, when alloyed with it makes it burn with great 
brilliancy, because there is formed a silicate of aluminium." 

While the above observations are in the main true, yet it is 
now well known that objects made of commercial aluminium do 
after a long exposure become coated with a very thin film, which 
gives the surface a "dead" appearance. The coating is very 
similar in appearance to that forming on zinc under the same 
circumstances. The oxidation, however, does not continue, for 
the film seems to be absolutely cdntinous and to protect the 
metal underneath from further oxidation. This coating can best 
be removed by very dilute acid (see Mourey's receipt, p. 6o), 
after which the surface can be burnished to its former brilliancy. 

The action of air and rain water together also slightly cor- 
rodes commercial aluminium, the sheet metal particularly 
showing after several months exposure small white spots of 
alumina wherever there was a speck or " piqfire." In no case, 
however, is this corrosion very strong or anything like as 
serious as the rusting of iron or corrosion of brass, copper, etc. 

It has also been found that at a high white-heat, especially 
at the heat of an electric furnace, aluminium burns with a strong 
light to alumina. It is quite probable that in this case it 
volatizes first, and it is the vapor which burns. During the 
operation of an electric furnace a white smoke formed of invisi- 
ble particles of alumina is thus formed and evolved from the 
furnace. Also, in melting aluminium, even the purest, it will 


be found that the surface seems bound and the aluminium 
restrained from flowing freely by a minute " sl^in " which may 
probably be a mixture of oxide with metal, or perhaps of oxides 
of foreign metals ; but, nevertheless, it is always present and is 
therefore indicative of oxidation taking place. It seems to pro- 
tect the metal beneath it perfectly, so that, once formed, it gets 
no thicker by continued heating. 

Wohler first discovered that when aluminium was in the ex- 
tremely attenuated form of a leaf it would burn brightly in air, 
and burn in oxygen with a brilliant bluish-light. It is also said 
that thin foil will burn in oxygen, being heated by wrapping it 
around a splinter of wood, and fine wire also burns like iron 
wire, but the combustion is not continuous, because the wire 
fuses too quickly. The alumina resulting is quite insoluble in 
acids, and as hard as corundum. 

The aluminium powder, produced by pulverizing aluminium 
leaf, burns brilliantly when projected into a flame, and has dis- 
placed, to a large extent, magnesium powder for making a flash 
Hght, because it is cheaper, gives a more strongly actinic light, 
and leaves no unpleasant smoke or fumes. For this purpose it 
is not used pure, but mixed with various chemicals, for the 
composition of which see the chapter on "Uses of Aluminium." 

Action of Water. 

Deville : " Water has no action on aluminium, either at ordi- 
nary temperatures or at 100°, or at a red heat bordering on the 
fusing point of the metal. I boiled a fine wire in water for half 
an hour and it lost not a particle in weight. The same wire was 
put in a glass-tube heated to redness by an alcohol lamp and 
traversed by a current of steam, but after several hours it had 
not lost its polish, and had the same weight. To obtain any 
sensible action it is necessary to operate at the highest heat of 
a reverberatory furnace — a white heat. Even then the oxida- 
tion is so feeble that it develops only in spots, producing almost 
inappreciable quantities of alumina. This slight alteration and 
■the analogies of the metal allow us to admit that it decomposes 


water, but very feebly. If, however, metal produced by M. 
Rose's method is used, which is almost unavoidably contam- 
inated with slag composed of chlorides of aluminium and 
sodium, the former, in presence of water, plays the part of an 
acid towards aluminium, disengaging hydrogen with the forma- 
tion of a subchlorhydrate of alumina, whose composition is not 
known, and which is soluble in water. When the metal thus 
tarnishes in water one may be sure to find chlorine in the water 
on testing it with nitrate of silver." 

Aluminium leaf, however, will slowly decompose water at 
ioo°. Hydrogen is slowly evolved, the leaf loses its brilliancy, 
becomes discolored, and, after some hours, translucent. It is 
eventually entirely converted into gelatinous hydrated alumina. 

A. Ditte * explains the action of water on aluminium as fol- 
lows : Water can only act on aluminium by producing hydro- 
gen gas and alumina, both of which deposit on the metal and 
cover it with a thin, protective coating. If the conditions are 
such that this coating is removed, the action becomes manifest. 
Boiling, for instance, removes the hydrogen, and if chloride, 
sulphate or nitrate of aluminium be present in the solution to 
remove the alumina, the action goes on, a basic salt being 
formed, and continues until a sub-salt is formed which is diffi- 
cultly soluble or insoluble, and covers the metal with a new, 
impermeable coating. Traces of these acids in the boiling 
water would lead to the same result. 

Action of Hydrogen Sulphide and Sulphur. 

Deville : " Sulphuretted hydrogen exercises no action on 
aluminium, as may be proved by leaving the metal in an 
aqueous solution of the gas. In these circumstances almost all 
the metals, and especially silver, blacken with great rapidity. 
Sulph-hydrate of ammonia may be evaporated on an alumin- 
ium leaf, leaving on the metal only a deposit of sulphur, which 
the least heat drives away. 

* Ann. de Chimie et de Physique [6], Vol. 20, p. 404 (i8go). 


" Aluminium may be heated in a glass tube to a red heat in 
vapor of sulphur without altering the metal. This resistance is 
such that in melting together polysulphide of potassium and 
some aluminium containing copper or iron, the latter are at- 
tacked without the aluminium being sensibly affected. Un- 
happily, this method of purification may not be employed 
because of the protection which aluminium exercises over 
foreign metals. Under the same circumstances gold and silver 
dissolve up very rapidly. However, at a high temperature I 
have observed that ,it combines directly with sulphur to give 
aluminium sulphide. These properties varying so much with 
the temperafure from one of the special characteristics of the 
metal and its alloys." 

Margottet states that hydrogen sulphide is without action on 
aluminium, as also are the sulphides of iron, copper, or zinc. 
Aluminium is said to decompose silver sulphide, AgjS, setting 
the sulphur, however, at liberty, and alloying with the silver. 
In regard to its indifference to the first mentioned sulphides, 
this would give inferential evidence that the reverse operation, 
i. e., the action of iron, copper, or zinc on aluminium sulphide, 
would be possible, as will be seen later to be apparently 
established by direct experiment. As to the action of sul- 
phuretted hydrogen, the author has a different experience to 
quote. On passing a stream of that gas into commercial alu- 
minium melted at a red heat, little explosive puffs were heard, 
accompanied by a yellow light, while the dross formed on the 
surface, when cooled, evolved sulphuretted hydrogen briskly 
when dropped into water, and gave every indication of contain- 
ing aluminium sulphide. It could not have been silicon sul- 
phide, for the metal contained as large a percentage of silicon 
after treatment as before. Hydrogen sulphide is also absorbed 
in large quantity by molten aluminium, and mostly evolved just 
as the metal is about to set. Some of the gas is entangled in 
the solidifying metal, forming and fiUing numerous cavities or 

88 aluminium. 

Sulphuric Acid. 

Deville : " Sulphuric acid, diluted in the proportion most 
suitable for attacking the metals which decompose water, has 
no action on aluminium ; and contact with a foreign metal does 
not help, as with zinc, the solution of the metal, according to 
M. de la Rive. This singular fact tends to remove aluminium 
considerably from those metals. To establish it better, I left 
for several months some globules weighing only a few milli- 
grammes in contact with the weak acid, and they showed no 
visible alteration ; however, the acid gave a faint precipitate 
when neutralized with aqua ammonia." 

It is a fact that dilute or concentrated sulphuric acid acts 
very feebly on pure aluminium in the cold, but on being heated 
they both attack it, disengaging sulphurous acid gas. Impure 
metal is attacked more easily than pure metal, the presence of 
silicon giving rise to the formation of silicon hydride, which 
communicates to the hydrogen set free a tainted odor. 

Ditte* describes the action of sulphuric acid as follows : A 
2.5 per cent, solution acts very slowly at first, if cold, but after 
several hours the air condensed on the aluminium is removed 
and the metal is slowly dissolved, evolving hydrogen. The 
bubbles of hydrogen, however, protect the surface from further 
attack, but anything which breaks up or removes this gaseous 
envelope hastens the action of the acid. Certain solutions of 
metallic chlorides which aluminium reduces act in this way. 
If, for instance, we add a few drops of platinic chloride to the 
above acid, platinum is reduced on the aluminium, the surface 
is roughened, causing the gas to escape quicker, and the action 
is much more rapid. Traces of chloride of gold, mercury or 
copper have the same effect. After some time, the action of 
the acid is stopped by the formation of an insoluble basic salt, 
having the formula 4AI2O3.3SO3.18H2O, which deposits on the 
metal and protects it from further action. Even very dilute 
sulphuric acid attacks aluminium if it is prevented from cover- 
ing itself with the layer of gas, as by boiling. 

♦Comptes Rendus, ex., pp. 583, 782 (1890). 














It appears to the writer that similar reasoning explains why 
impure aluminium is attacked more rapidly. Aluminium being 
electro-positive towards the impurities it contains, it follows 
that when the aluminium is at all attacked by the acid the gas 
will be disengaged on the particles of the impurity, and not on 
the aluminium itself ; thus the pure aluminium will never get 
the benefit of the gaseous protection, and the action proceeds 

G. A. Le Roy* tested four specimens of commercial alu- 
minium, whose analysis was as follows : 


Aluminium 98.28 

Iron 1 .5o 

Silicon o. 1 2 

Specimens A and B were made by the sodium process at 
Nanterre ; C and D by another process, not named. The metal 
was in sheets and was cut to a definite size, cleaned in soda, 
washed in alcohol, dried, weighed, and then put in the acid. 
Afterwards they were withdrawn, washed with water, dipped in 
alcohol, dried and weighed. The results are expressed as the 
weight in grammes dissolved during twelve hours from a sur- 
face of one square metre : 

Quality of Acid. Gravity. 

Pure 1.842 

Commercial i .842 

Pure 1. 711 

Commercial 1. 711 

Pure 1.580 

Pure 1.263 

Pure 1.842 

Commercial i .842 

From these tests Le Roy concludes that it is impracticable 
to think of using aluminium for the different apparatus, such as 
pans, pumps, tank linings, etc., used in the manufacture or 
handling of sulphuric acid. 

* Le Moniteur Industriel, Oct. 29, 1891. 



Loss in grammes per square metre in t2 hours^ 



































.. .. 
















While these tests show that the aluminium is certainly dis- 
solved, yet they also prove that the action is very slow. If 
these acids, for instance, were contained in an aluminium vessel 
I millimetre in thickness (0.04 inch), we can easily calculate 
from the above figures that the concentrated acid could rest in 
it cold from 60 to 90 days before eating through ; the 30° acid, 
which is yet strong, would take 290 to 560 days to get through 
it; while even the hot concentrated acid would not get through 
in less than 5 days' constant action. We can infer that cold, 
dilute acid would probably take several years in producing the 
same efTect. 

Nitric Acid. 

Deville : " Nitric acid, weak or concentrated, does not act on 
aluminium at the ordinary temperature. In boiling acid, solu- 
tion takes place, but with such slowness that I had to give up 
this mode of dissolving the metal in my analyses. By cooling 
the solution all action ceases. On account of this property, 
M. Hulot has obtained good results on substituting aluminium 
for platinum in the Grove battery." 

Ditte* describes the action of nitric acid similarly to that 
of sulphuric. Cold 3 per cent, acid acts very slowly, 6 
per cent, quicker but slower than sulphuric acid of the 
same strength. When the plate becomes rough or mat by 
the action, the gas escapes more freely from the surface 
and the attack is more rapid. The gas evolved contains 
no hydrogen, but is mostly nitrous-oxide, similarly to the 
action on zinc, with a little ammonia gas. The reaction does 
not stop with the formation of the neutral nitrate, but this is 
acted on by the aluminium, forming a white, granular precipi- 
tate of a basic salt having the formula : 6AI2O3.3N2O5.30H2O. 

While cold acid thus attacks the metal very slowly, the hot 
concentrated acid at 100° attacks it violently. 

The amount of the action of cold nitric acid has been mea- 
sured by Le Roy, Lunge, and the writer, as follows : 

* Vide, p. 88. 



Le Roy, pure, concentrated i5"-20° 

" commercial, concentrated ! " 

" " 52 per cent ! " 

Lunge, pure concentrated. 
" " 65 per cent. . . 

" " 32 per cent. . . 

Richards, pure, concentrated. 


B * 

6 5!" 

cJ rt fi 

t, jj -rH 

^ t/> i^ u 







The figure last given includes the loss sustained in polishing 
the metal to its original appearance before immersion. It re- 
sults from these tests that a sheet of aluminium one millimetre 
thick, if left standing in concentrated nitric acid, would sustain 
constant immersion at ordinary temperatures for 3600 days 
(Lunge), 43 days (Le Roy), or 14 days (Richards), if taken 
out every day and polished bright. 

Concerning the use of aluminium in the Grove battery, which 
is not a constant battery, but only used to give a strong cur- 
rent for a short time, aluminium sheet one millimetre thick 
will last a long time and cost only a fraction as much as thin 

Hydrochloric Acid. 

This acid attacks commercial aluminium rapidly, quicker 
when hot than when cold, and the concentrated quicker than 
the dilute. The very pure metal, however, is attacked slowly, 
showing that, as with sulphuric acid, the galvanic action of the 
impurities keeping the aluminium free from a protecting layer 
•of gas, has a great influence on the resistance to acids. Gase- 
ous hydrochloric acid attacks the metal even at a very low 

The presence of silicon, particularly, increases the facility 


with which the metal is attacked. When silicon is present, the 
graphitic silicon remains behind as a black residue, while the 
combined silicon partly forms silica and partly escapes as sili- 
con hydride (SiHi), having a very disagreeable smell. When 
the amount of combined silicon is small, almost all of it may 
thus escape. In analyzing aluminium, this loss is prevented by 
keeping the acid well oxidized by bromine water, which pre- 
vents the formation of the gas. On evaporating to dryness,, 
after solution, and taking up with water, the graphitoidal sili- 
con remains as a black crystalline residue, while the combined 
silicon has all been changed to white silica. These can be 
separated by washing with hydrofluoric acid, which dissolves 
the silica but leaves the silicon unattacked. 

If a small amount of hydrochloric acid is present in a mix- 
ture of acids, it leads the attack on the metal, and the other 
acids form aluminium salts by reacting on the aluminium 
chloride formed. Hydrobromic, hydriodic and hydrofluoric 
acids act similarly to hydrochloric. Aluminium fluoride, how- 
ever, is not so soluble in water as the other halogen salts, and 
the attack by hydrofluoric acid is therefore hindered by the 
formation of an insoluble coating on the aluminium when the 
solution passes a certain degree of concentration. 

In an experiment by the writer, best commercial rolled alu- 
minium was put into cold, dilute (3 per cent.) hydrochloric 
acid. It was some time before any action was observable, and 
then it was very slow. It lost, after cleaning, at the rate of 5S 
grammes per square metre of surface per day, which was only 
a little over half as much as was dissolved by cold, concentrated 
nitric acid in the same time. A piece of the same metal, how- 
ever, alloyed with three per cent, of nickel was rapidly attacked 
under these conditions, and over thirty times as much of it was 
dissolved in a given time, thus showing strikingly the influence 
of a small amount of impurity. 

Organic Acids, Vinegar, Etc. 
Deville : " Weak acetic acid acts on aluminium in the same 


way as sulphuric acid, i. e., in an inappreciable degree or with 
extreme slowness. I used for the experiment acid diluted to 
the strength of strongest vinegar. M. Paul Morin left a plaque 
of the metal a long time in wine which contained tartaric acid 
in excess and acetic acid, and found the action on it quite in- 
appreciable. The action of a mixture of acetic acid and com 
mon salt in solution in pure water on pure aluminium is quite 
different, for the acetic acid replaces a portion of the chlorine 
existing in the sodium chloride, rendering it free. However, 
this action is very slow, especially if the aluminium is pure." 

" The practical results flowing from these observations deserve 
to be clearly defined, because of the applications which may be 
made of aluminium to culinary vessels. I have observed that 
the tin so often used and which each day is put in contact with 
common salt and vinegar, is attacked much more rapidly than 
aluminium under the same circumstances. Although the salts 
of tin are very poisonous, and their action on the economy far 
from being negligible, the presence of tin in our food passes un- 
perceived because of its minute quantity. Under the same 
circumstances aluminium dissolves in less quantity; the acetate 
of aluminium formed resolves itself on boiling into insoluble 
alumina or an insoluble sub-acetate, having no more taste or 
action on the body than clay itself. It is for that reason and 
because it is known that the salts of the metal have no appre- 
ciable action on the body, that aluminium may be considered 
as an absolutely harmless metal." 

The low price of aluminium has within the last two years 
rendered possible this application to culinary utensils foreseen 
by Deville, and has led to several investigations as to its resist- 
ance to the acids and salts liable to be met with in food. 

Two German pharmacists, Lubbert and Roscher, conducted 
tests of this kind in 1891, but their results were only qualitative, 
not quantitative, and they made the mistake of using thin 
aluminium foil, which averages only o.ooi inch thick, instead 
of heavier sheet. The result was that they found their thin foil 
totally dissolved by two days immersion in oxalic acid (1,5 


and lO per cent.), gallic acid (2 per cent.), and corrosive sub- 
limate (i per cent.) ; and by four days constant immersion in 
I, 5 and 10 per cent, solutions of formic, acetic, butyric, lactic,, 
tartaric and citric acids, pure oleic acid, 10 per cent, solutions 
of palmitic or stearic acids, in 5 per cent, solutions of carbolic 
acid, 4 per cent, solution of boric acid, and in pure red wine, 
white wine, coffee and tea. 

It has been found, however, that because of its thinness alu- 
minium foil does not resist corrosion as does the heavier sheet- 
metal, and while the above results may be strictly true for foil 
0.00 1 inch thick, yet ordinary sheet aluminium is not attacked 
to anything like such an extent. Culinary utensils are made of 
sheet at least i millimetre (0.04 inch) thick, and tests of mate- 
rial of this thickness made by subsequent investigators have 
given very different results. 

Balland* conducted tests for several months, and found that 
air, water, wine, beer, cider, coffee, milk, oil, butter, fat, urine, 
saliva and damp earth have less action on aluminium than on 
iron, copper, lead, zinc, or tin. Vinegar and salt together at- 
tack it, but so slightly as not to prevent its use for cooking; a 
sheet only lost 1.3 per cent, of its weight in vinegar, and 0.2 
per cent, of its weight in a 5 per cent, salt solution after four 
months' immersion. 

Ruppf states that he has kept different kinds of liquid and 
semi-solid food in aluminium vessels for 4 to 28 days, at the 
ordinary temperature, and found the metal unaltered. The 
metal used contained 0.30 per cent, of iron, 0.08 silicon and 
99.66 aluminum. 

Prof. Geo. Lunge, | of Zurich, has made a most careful and 
systematic investigation of this subject. Assisted by Ernst 
Schmid, he made the following series of tests ; ordinary com- 
mercial sheet aluminum from Neuhausen was used. It ana- 
lyzed : 

* Comptes Rendus, 1892, Vol. 114, p. 1536. 

t Chronique Industrielle, June 12, 1892. 

I Zeitschrift fiir angewandte Chemie, January, 1892. 


Aluminium 99.20 per cent, (by difference) 

Combined silicon 0.44 " 

Graphitic " o.i I " 

Iron 0.25 " 

Copper trace. 

The sheet was i minimetre thick, and was cut into strips of 
a fixed size. These were first carefully cleaned by caustic soda 
and sulphuric acid, and then immersed six days in the liquids 
at the ordinary temperature, each liquid being tested twice to 
guard againt mistakes. The following are their results, ex- 
pressed as the weight dissolved per day in milligrammes per 
square metre of surface exposed : 

Loss of weight per day in 
milligrammes per square 

Liquid. i metre of surface. 

Ordinary claret 47.3 

" hock 51.2 

Brandy 18.0 

Pure 50 per cent, alcohol 10.2 

Tartaric acid, 5 per cent, solution 28.2 

" '■ I " " 43.0 

Acetic " S " " .64.2 

" " I " " 73.0 

Citric " 5 " " 35-8 

" I " " 31-7 

Lactic " 5 " " ■■■■■ 79-5 

Butyric " 5 " " 21.8 

Carbolic " 5 " " 3-8 

" " 1 " •• 8.2 

Boric " 4 " " 29.5 

Salicylic " I4 " " 105.8 

Coffee (poured hot) 8.3 

Tea " 00 

Beer o-o 

The aluminium strips used showed outward evidences of 
corrosion in only a few places ; only in the case of the salicylic 
acid solution had the metal lost its bright surface and become 
dull. In the case of brandy and alcohol, where the quantitative 
action was very slight, the surface of the aluminium showed a 
few fungus-like excresences, probably formed by alumina, 
caused by accidental flaws in the sheet. In the case of a wine 


rich in tannin, these spots are liable to be darkened by the in- 
fluence of the trace of iron dissolved from the aluminium, form- 
ing tannate of iron, but the amount thus formed is infinitesmal. 
The weights dissolved are so extremely small, that it may be 
calculated that a canteen holding a litre could be kept full of 
the strong acetic acid for 55 years before losing half its weight. 

Lunge and Schmid conclude that, " the action of coffee, tea 
and beer is practically zero ; that of acids and acid liquids is 
more pronounced, but in the worst case too slight to cause any 
alarm whatever. Nor is there the slightest danger of any in- 
jurious action on the human body by such traces of aluminium 
compounds, seeing that our food contains very much more than 
these ; in fact, they could not act injuriously unless quantities 
hundreds of times larger were regularly entering the stomach." 

The only criticism we have to make of this valuable investiga- 
tion is that the solutions should have been tested hot, even 
boiling, as well as cold. While it is perfectly true, as remarked 
by a Washington lady, that " we hope never to have occasion 
to serve our families with poisoned soups or salicylic or boracic 
acid stews," yet if Lunge had made the hot-liquid tests we 
should, at least, have known the worst that could possibly hap- 
pen to the aluminium culinary utensils. 

Happily, actual experience has settled all these questions. 
An aluminium sauce-pan used in the writer's family constantly 
for two years, which has been used for boiling milk, cooking 
oatmeal and vegetables, stewing meats and preserving several 
kinds of fruits, lost less than one-quarter ounce in the whole two 
years, an average of less than 15 milligrammes per day per 
square metre of surface exposed, including frequent cleaning. 
I have calculated that at this rate the utensil will become a fam- 
ily heirloom unless accident overtakes it, for it could stand this 
rate of wear for over a century before losing half its weight. 
Other experiences equally as striking could be cited, and no 
better proof of the suitability of aluminium for culinary utensils 
could be wished. The German Minister of War, after thorough 
experiments by an army commission, has adopted aluminium 


canteens and drinking cups for the soldiers, and entire sets of 
cooking utensils for use in the officers' quarters. It is the 
writer's opinion that within a short time the larger part of all 
the aluminium made will be needed to supply the great demand 
which will arise for these culinary utensils. 

Sodium Chloride. 

Common salt does not, when molten, corrode aluminium, so 
that it forms a good flux to melt the metal with. It does not 
possess the property, like fluorspar, of dissolving alumina, but 
acts simply by its fluidity to give a clean melt. 

A solution of common salt acts very feebly on aluminium. 
A strip of rolled aluminium immersed in a three per cent, solu- 
tion at 27° C. lost weight at the rate of 394 milligrammes per 
day per square metre of surface (Hunt). The writer found a 
corresponding loss of 400 milligrammes per day on sheet im- 
mersed in a strong brine kept at 65° C. (150° F.) At that 
rate it would take nine years' constant immersion to entirely 
dissolve a sheet one millimetre thick. 

When an acid is present with the salt, the action is stronger 
than when either is alone. Mr. Hunt found that when two 
per cent, of acetic acid was added to the salt solution previously 
mentioned, the corrosion was increased to 1738 milligrammes 
per day. This solution will fairly represent the extreme con- 
ditions to which aluminium will be subjected in domestic culi- 
nary operations, and shows that the corrosion is so slight as to 
be of no practical consequence, being much less than copper, 
tin-plate or iron would sufifer under similar conditions. 

Sea-water corrodes aluminium slightly, but much less than 
iron or steel, under similar conditions. Strips of aluminium 
on the sides of a wooden sailing vessel lost less than 100 milli- 
grammes per square metre of surface during six months' ex- 
posure (Hunt). Messrs. Yarrow & Co., on the Thames, 
having in view the building of an aluminium torpedo boat for 
the French government, took two plates of aluminium stiffened 
by alloying with six per cent, of copper, and after weighing 


accurately secured them on the sides of a wooden, coppered 
sailing vessel, the copper being removed to make place for 
the aluminium. This ship made a voyage round the world ; 
then the aluminium plates were removed, weighed, and found 
to have suffered no appreciable loss. As a consequence of this 
result, Messrs. Yarrow proceeded with the construction of the 
boat, which is the largest vessel yet built of aluminium, having 
a length of sixty feet, beam nine feet three inches and a maxi- 
mum speed of twenty-two knots (twenty-five and one-half 
miles) per hour. 

It is to be recommended, however, that aluminium hulls in- 
tended for salt water should be painted, in order to reduce any 
possible corrosion to a minimum. 

Organic Secretions. 

These act only slightly on aluminium, the degree of corrosion 
being proportioned mostly to the amount of sodium chloride 
present in the secretion and its degree of acidity, or, particu- 
larly, of alkalinity. The saliva acts on it so slightly that alu- 
minium dental plates may be worn for many years without any 
appreciable corrosion. M. Charriere, a French physician, was 
the first to make a small tube of it for a person on whom trache- 
otomy had been performed. He found it to remain almost 
unaltered by contact with purulent matter ; after a long time a 
little alumina was formed on it, hardly enough to be visible. 
The use of aluminium for suture wire is also highly recom- 
mended ; many instruments for physician's use are being made 
of aluminium, its great advantages being lightness, incorrodibil- 
ity, and being so easily cleansed, particularly by antiseptic 
solutions. The tarnishing of polished aluminium articles when 
constantly handled is due to the perspiration, which contains 
about two per cent, of sodium chloride and about an equal 
quantity of organic acids. Its action is not very strong, yet 
sufficient to spoil a high polish and give a visible tarnish, as 
would indeed happen to almost any other metal. If, however, 
the perspiration is profuse and acts on the aluminium con- 


stantly while warm, a deeper corrosion results. I have seen an 
aluminium ankle-stiffener, of metal about one millimetre thick, 
which was worn in the heel of a boot for about a year, and was 
then eaten nearly through by the corrosion due to perspiration 
a little above the ankle. The shoe lining which covered it had 
served to keep it constantly in contact with the perspiration, 
while the temperature may have been about 85° C. This is, 
however, an extreme case, and it would be interesting to know 
whether steel or very stifif leather would have lasted as long_ 
under the same circumstances. 

Caustic Alkalies. 

Deville : " Solutions of caustic potash or caustic soda ict' 
with great energy on aluminium, transforming it into aluminate 
of potash or soda, setting free hydrogen. However, it is not 
attacked by caustic potash or soda in fusion ; one may, in fact, 
drop a globule of the pure metal into melted caustic soda raised 
almost to red heat in a silver vessel, without observing the 
least disengagement of hydrogen. Silicon, on the contrary, 
dissolves with great energy under the same circumstances. I 
have employed melted caustic soda to clean siliceous alumin- 
ium. The piece is dipped into the bath kept almost at red 
heat. At the moment of immersion several bubbles of hydro- 
gen disengage from the metallic surface, and when they have 
disappeared all the silicon of the superficial layer of aluminium 
has been dissolved. It only remains to wash well with water 
and dip it into nitric acid, when the aluminium takes a beautiful 

Mallet found that the purest aluminium resists the caustic 
alkaline solution better than the commercial metal. I have- 
noticed the same fact in comparing aluminium of different 
degrees of purity; best commercial metal withstood the action 
of cold, dilute caustic potash solution seven times better than 
when it contained three per cent, of copper, and seventy times 
better than when two per cent, of copper and one per cent, of 
zinc were present. 


Prof. A. Stutzer,* of Bonn, finds that the metal made by the 
electrolytic processes dissolves much slower than that made by 
the former sodium processes. He ascribes this to the latter 
metal containing nearly one per cent, of sodium. This, how- 
ever, can hardly be the case, because metal made by the Deville 
process rarely contained more than a trace of sodium. The true 
reason must be simply the greater purity of the metal now be- 
ing made by electrolysis, which averages over 99 per cent, pure 
for best quality metal, while the Deville metal was rarely over 
98 per cent. pure. 


Aqua ammonia acts slowly on aluminium, producing a little 
alumina, part of which remains dissolved. Ammonia gas does • 
not appear to act on the metal. 

Lime Water. 

This acts rapidly on aluminium at first, but the resulting 
calcium aluminate is insoluble in water, and is, therefore, pre- 
cipitated on the metal and quickly protects it from further 
action. If this coating is scraped off the action is repeated. 

Solutions of Metallic Salts. 
Deville : " The action of any salt whatever on aluminium may 
be easily deduced from the action of its acids on that metal. ' 
We may, therefore, predict that in acid solutions of sulphates 
and nitrates aluminium will precipitate no metal, not even 
silver, as Wohler has observed. But the hydrochloric solutions 
of the same metal will be precipitated, as MM. Tissier have 
shown. Likewise, in alkaline solutions, silver, lead, and metals 
high in the classification of the elements are precipitated. It 
may be concluded from this that to deposit aluminium on other 
metals by means of the battery, it is always necessary to use 
acid solutions from which hydrochloric acid, free or combined, 
.should be absent. For similar reasons the alkaline solutions of 

* Zeitschr. fiir angewandte Chemie, 1890, No. 23. 


the same metals cannot be employed, although they give such 
good results in plating common metals with gold and silver. It 
is because of these curious properties that gilding and silvering 
aluminium are so difficult." 

These conclusions by Deville are confirmed only when using 
pure aluminium; the impure metal, containing iron, silicon, or 
perhaps sodium, may produce very slight precipitates in cases 
where pure aluminium would produce none. Some observers 
have noted different results in some cases even when using alu- 
minium free from these impurities. We will therefore take up 
these cases and consider them separately. 

Mercury. — * Aluminium decomposes solutions of mercuric 
chloride, cyanide or nitrate, mercury separating out first, then 
forming an amalgam with the aluminium which is immediately 
decomposed by the water, the result being alumina and mer- 
cury. From an alcoholic solution of mercurous chloride the 
mercury is precipitated more quickly at a gentle heat. A solu- 
tion of mercurous iodide with potassium iodide is also reduced 
in like manner. 

Copper. — * From solution of copper sulphate or nitrate, alu- 
minium separates out copper only after two days' standing, as 
either dendrites or octahedra ; from the nitrate it also precipi- 
tates a green, insoluble basic salt. Copper is precipitated 
immediately from a solution of cupric chloride; but slower 
from the solution of copper acetate. The sulphate or nitrate 
solutions behave similarly if potassium chloride is also present, 
and the precipitation is complete in presence of excess of alu- 

Silver. — *From a nitrate solution, feebly acid or neutral, 
aluminium precipitates silver in dendrites, the separation only 
beginning after six hours' standing. From an ammoniacal 
solution of silver chloride or chromate, aluminium precipitates 
the silver immediately as a crystalline powder. Cossa con- 
firms the statement as to the nitrate solution. 

Lead. — * From nitrate or acetate solution the lead is slowly 

* Dr. Mierzinski. 


precipitated in crystals ; an alkaline solution of lead chromate 
gives precipitates of lead and chromic oxide. 

Zinc. — * An alkaline solution of zinc salt is readily decom- 
posed and zinc precipitated. 

Margottet states that all metallic chlorides excepting those 
of potassium or sodium are reduced from solution. This state- 
ment can hardly include chlorides of magnesium or lithium, 
since magnesium precipitates alumina from solutions of alu- 
minium salts. Alkaline or ammoniacal solutions are more 
easily decomposed than acid solutions ; in alkaline solutions 
the cause being the facility with which aluminates of the alka- 
lies are formed. 

Alkaline Chlorides. — A solution of sodium or potassium 
chlorides is not appreciably affected by pure aluminium, either 
cold or warm. However, rather impure aluminium which was 
packed in a case with saw-dust and kept wet with sea-water for 
two weeks was corroded ; whether the result would have been 
the same without the presence of the saw-dust or with purer 
aluminium, I cannot say. 

Aluminium Salts. — It is a curious fact that a solution of alu- 
minium chloride will attack aluminium, forming a basic-chloride 
with evolution of hydrogen. A solution of alum does not attack 
aluminium, but if sodium chloride is added it is dissolved with 
evolution of hydrogen. It is interesting to note that while 
neither of these salts alone attacks aluminium, the mixture of 
the two does. 

Fluorspar on being melted gives off a little hydroflouric acid 
vapor, caused by the action of the hydroscopic moisture in it. 
This vapor will attack aluminium, forming fluoride. Aside 
from this slight action, the molten fluorspar has no action on 
the aluminium. It has, however, the property of dissolving 
a small proportion (about 2 per cent.) of alumina, and this 
action causes it to be a very efficient flux, because small 

* A. Cossa, Bull, de la Soc. Chim., 1870, p. 199. 


globules of aluminium are usually encrusted with a thin film 
of alumina which prevents them from running together to a but- 
ton. The fluorspar, by dissolving this coating and affording a 
fluid mass, fluxes the separated globules together. It is often 
used in connection with cryolite and common salt. 


Melted cryolite is a very good flux for aluminium, for the 
same reason as described under fluorspar. It can, however, 
dissolve over 20 per cent, of alumina, and is therefore that 
much more efficient as a fluxing agent. At a temperature 
above the melting point of copper it attacks the aluminium 
itself, if it is put into it finely divided, but the metal en masse 
is not sensibly attacked. The result of this attack must be the 
formation of a sub-fluoride, possibly AIF2. 

Silicates and Borates. 

Neither of these classes of compounds can be used as fluxes 
or slags in working aluminium, since they both rapidly corrode 
the metal. Deville had little difficulty in decomposing these 
salts so completely with metallic aluminium that he isolated 
silicon and boron. If aluminium is melted in an ordinary 
glass vessel it attacks it, setting free silicon from silica, forming 
an aluminate with the alkali present and an alloy with the sili- 
con set free. Aluminium melted under borax is rapidly dis- 
solved, an aluminium borate being formed. It is thus seen 
that the common metallurgic slags are altogether excluded 
from the manufacture of aluminium, and also that it is an im- 
possibility to manufacture pure aluminium direct from minerals 
containing silica. The result of reducing kaolin, for instance, 
would inevitably be highly siliceous aluminium, from which 
there is no known method of separating out pure aluminium. 
These facts also point to the importance of keeping molten 
aluminium from contact with the ordinary siliceous refractory 
materials in working it. Even the best quality plumbago 
crucibles will increase the percentage of silicon in commercial 


aluminium at least 0.25 per cent, at one melting. Magnesia 
brick linings and magnesia-lined crucibles should be used for 
melting it in. 


Deville : " Aluminium may be melted in nitre without under- 
going the least alteration, the two materials rest in contact 
without reacting even at a red heat, at which temperature the 
salt is plainly decomposed, disengaging oxygen actively. But 
if the heat is pushed to the point where nitrogen itself is disen- 
gaged, there the nitre becomes potassa, a new affinity becomes 
manifest, and the phenomena change. The metal then com- 
bines rapidly with the potassa to give aluminate of potash. 
The accompanying phenomenon of flagration often indicates a 
very energetic reaction. Aluminium is continually melted with 
nitre at a red heat to purify it by the oxygen disengaged, with- 
out any fear of loss. But it is necessary to be very careful in 
doing it in an earthen crucible. The silica of the crucible is 
dissolved by the nitre, the glass thus formed is decomposed by 
the aluminium, and the silicide of aluminium formed is then 
very oxidizable, especially in the presence of alkalies. The 
purification by nitre ought to be made in an iron crucible well 
oxidized by nitre inside." 

If finely divided aluminium is mixed with nitre and brought 
to a red heat, the metal is oxidized with the production of a 
fine blue flame. (Mierzinski.) Nitre is also used sometimes 
in the composition of aluminium flash-light powders. 

Alkaline Sulphates and Carbonates. 
Tissier : "Only 2.65 grammes of aluminium introduced into 
melted red-hot sodium sulphate (NajSOj) decomposed that 
salt with such intensity that the crucible was broken into a 
thousand pieces, and the door of the furnace blown to a dis- 
tance. Heated to redness with alkaline carbonate, the alu- 
minium was slowly oxidized at the expense of the carbonic 
acid, carbon was set free, and an aluminate formed. The re- 
action takes place without deflagration." 

chemical properties of aluminium. i05 

Metallic Oxides. 

Tissier Brothers made a series of experiments on the action 
of aluminium on metallic oxides. Aluminium leaf was care- 
fully mixed with the oxide, the mixture placed in a small por- 
celain capsule and heated in a small earthen crucible, which 
served as a muffle. The results were as follows : 

Manganese dioxide. — No reaction. 

Zinc oxide. — No reaction even at white heat. 

Ferric oxide. — By heating to white heat one equivalent of 
ferric oxide and three of aluminium, the reaction took place 
with detonation, and by heating sufficiently we obtained a 
metallic button, well melted, containing 69.3 per cent, of iron 
and 30.7 per cent, of aluminium. Its composition corresponds 
very nearly to the formula AlFe. It would thus appear that 
the decomposition of ferric oxide will not pass the limit where 
the quantity of iron reduced is sufficient to form with the alu- 
minium the alloy AlFe. 

Lead oxide. — We mixed two equivalents of litharge with one 
of aluminium, and heated the mixture slowly up to white heat, 
when the latter reacted on the litharge with such intensity as 
to produce a strong detonation. We made an experiment with 
fifty grammes of litharge and 2.9 grammes of aluminium leaf, 
when the crucible was broken to pieces and the doors of the 
furnace blown off. 

Copper oxide. — Three grammes of black oxide of copper 
mixed with 1.03 grammes of aluminium detonated, producing 
a strong explosion, when the heat reached whiteness. 

Beketoff"* reduced baryta (BaO) with metallic aluminium in 
excess, and obtained alloys of aluminium and barium contain- 
ing in one case twenty-four per cent., in another thirty-three 
per cent, of barium. 

If any of the metallic oxides are fluxed, as by being dis- 
solved in cryolite, metallic aluminium reduces them to metal at 
once. Messrs. Green and Wahl have patented this method of 

* Bull, de la Soc. Chimique, 1887, p. 22. 


producing pure metallic manganese, which is at present being 
put into operation in Philadelphia on a commercial scale. 

Miscellaneous Agents. 

Phosphate of lime. — Tissier Brothers heated to whiteness a 
mixture of calcium phosphate with aluminium leaf, without the 
metal losing its metallic appearance or any reaction being 

Hydrogen. — This gas appears to have no action on alumin- 
ium, except to be dissolved in it in a moderately large quantity. 

Chlorine. — Gaseous chlorine attacks the metal rapidly. Alu- 
minium foil heated in an atmosphere of chlorine takes fire and 
burns with a vivid light. 

Bromine, iodine, act similarly to chlorine. 

Fluorine. — Fluorine gas forms a thin coating of fluoride which 
prevents a deeper attack. If the metal is at a dark-red heat, a 
violent incandescence ensues, and the attack is very energetic. 
Under the microscope the residue is seen to be globules of alu- 
minium covered with uncrystallized fluoride. 

Silver chloride. — Fused silver chloride is decomposed by 
aluminium, the liberated silver as well as the excess of alumin- 
ium being melted by the heat of the reaction. 

Mercurous chloride. — If vapors of mercurous chloride are 
passed through a tube in which some hot aluminium is placed, 
mercury is separated out, aluminium chloride deposits in the 
cooler part of the tube, and the aluminium is melted by the 
heat developed. 

Carbonic oxide. — Aluminium acts on this gas, forming alu- 
mina and setting free carbon. Prof. Arnold proved this by 
taking molten steel of a known content of carbon, and after 
adding to it several per cent, of aluminium, he passed a current 
of this gas through it for one hour. A sample of the metal 
subsequently analyzed showed one-third more carbon than 
before the treatment. This reaction is one of the causes of 
the advantage of adding aluminium to molten steel, as it thus 
jemoves one of the gases which form blow-holes in the cast- 



In this chapter we propose to note in rather condensed form 
the prominent characteristics of the various aluminium com- 
pounds, with an outline of the methods by which they can be 
produced, reserving for another chapter, however, the prepara- 
tion of those salts which are now being manufactured on a 
commercial scale for purposes of further treatment for alumin- 
ium. I do not propose this as a substitute for the various 
chemical treatises on this subject, but simply to add to the 
completeness of this work in order that a fair understanding of 
the other parts of the book may not be missed because data of 
this nature are not immediately at hand. Parts of this chapter 
are taken from M. Margottet's treatise on aluminium, in 
Fremy's Enclycopedie Chimique. 

General Considerations. 

Position of aluminium in the periodic classification of the ele- 
ments. — Mendeleeff places aluminium in his third family of 
elements, it being preceded by boron, and followed by the rare 
elements scandium, gallium, yttrium, indium, didymium, erbium 
and thallium. These elements all form oxides of the same 
general form, R2O3, and their other chemical compounds are 
■correspondingly similar. As we have zinc, cadmium and mer- 
cury as the analogues of magnesium in the second group, so we 
have gallium, indium and thallium as the corresponding ana- 
logues of aluminium in the third group. Unfortunately, these 
elements are so rare that we are not so familiar with their prop- 
erties as we could wish. We can notice, however, that as the 
atomic weight increases the specific gravity and the atomic 

(107 ) 


volume increase also, while the compounds formed by these 
metals become more easily decomposable. For instance : 

Element. Atomic Weight. Specific Gravity. Atomic Volume. 

Boron ii 2.5 4.5 

Aluminium 27 2.7 10 

Gallium 70 5.9 12 

Indium 113 7.4 14 

Thallium 204 11.8 17 

Structure of aluminium compounds. — Aluminium has until re- 
cently been regarded as a quadrivalent element, for the reason 
that no compound was known whose molecule contained less than 
two atoms of aluminium combined with six atoms of a mono- 
valent element. Thus, the oxide was AI2O3, the chloride AhCle,, 
etc. It was, therefore, supposed that in the molecule of every 
aluminium compound the two atoms of aluminium were held 
together by an exchange of one bond, and since each atom had 
three bonds left over, it was supposed that each single alumin- 
ium atom had four bonds or affinities. The composition of the 
oxide and chloride were therefore represented graphically as. 
follows : 

r O -, Cl cl 

= A1— A1 = Cl — Al — Al — Cl 

A dl 

The two atoms of aluminium, having six free affinities, were 
spoken of as a " double hexad-atom." 

All this reasoning, however, has broken down since the dis- 
covery of compounds containing only one atom of aluminium! 
in a molecule, and in which the aluminium is undoudtedly tri- 
valent; such are the organic salts A^CHa),, A1(C2H5)3,. 
A1(C5H702)3. In the latter salt, the vapor density corresponds 
to the formula given at only 45° above its melting point- 
Even aluminium chloride itself has been proven to have the 
formula AlCls when its vapor density is taken at a temperature 
above 800°. We are therefore constrained to drop the former 


ideas, and consider aluminium as tri-valent. The graphic 
formulae given above must then be written 

= A1 — O — Al = o Cl — Al — Cl 

The tri-valency of aluminium is also required by its position 
in the list of elements. Thus we have, taking the elements 
immediately preceding and following it : 

Atomic Weight. Valence in organic salts 

Sodium 23 I 

Magnesium 24 2 

Aluminium 27 3 

Silicon 28 4 

Phosphorus 31 3 

Sulphur 32 2 

Chlorine 353^ i 

Aluminium acts as a feeble base or a rather strong acid. As 
a base it has many analogies with iron in ferric compounds. 
For instance, the oxides AlzOs and FcjOg crystallize in the same 
iorms, rhombohedra, with almost exactly the same fundamen- 
tal angles, while the native hydrous oxides AI2O3.H2O and 
Fe^Os.HjO, both crystallize in the orthorhombic system with 
very similar angles and cleavages. As an acid, it forms com- 
pounds analogous to ferric and chromic acids. For instance : 

Aluminate of iron ^. FeAl^O,. 

Ferrate " " FeFejOj. 

Chromate " " FeCr^O^. 

These all crystallize in isometric octahedra, and are in every 
respect very similar minerals. The salts in which aluminium 
acts as an acid are called aluminates. They may be written 
as salts of aluminic acid, HjAljO^, but there are other alumin- 
ates known having a smaller proportion of alumina, and it is 
much more convenient to express the composition of all these 
aluminates by the dualistic method. Thus, there are known to 
exist (see Bayer's method of treating bauxite, next chapter.) 



the writing of which formulae by the rationalistic method would 
be very cumbersome. 

The normal or neutral salts of aluminium as a base may be 
regarded as the acid with its hydrogen replaced by alumin- 
ium. Thus : 

Acid. Salt. 

3HCI AlCls 

3QH,0, AlCQHsO,), 

3H,S0t A1,(S0J3 

HaPjO^ AIP2O4 

Aluminium, however, has a great tendency to form basic 
salts when these neutral salts are put in contact with an excess 
of aluminium or alumina. It is most convenient to write the 
formulae of these by the dualistic method. Thus we have : 

Normal sulphate, AI2O3.3SO3.16HJO. Basic sulphate, 2AI2O3.SO3.10H2O. 

Normal acetate, AI2O3.6C2H2O. Basic acetate, Al20s.2C2H.^0.2H20. 

While the writer has no intention of going back to the 
duaHstic theories which such formulae were first used to repre- 
sent, yet he takes the liberty of using these formulas to represent 
the empirical composition of these salts, because of their con- 
venience, and without attaching to them any theories as to the 
internal structure of the molecules. 

General methods of formation and properties. — Hydrated alu- 
mina, which has not been too strongly heated, dissolves in 
strong acids, forming salts which are mostly soluble in water. 
In the feebler acids and in all organic acids it is completely in- 
soluble. The salts of these latter acids are formed best by de- 
composing solution of aluminium sulphate with the barium or 
lead salt of the acid in question. Most aluminium salts are 
soluble in water and rather difficult to crystallize : the few in- 
soluble salts are white, gelatinous, and similar to the hydrate 


in appearance. In the neutral salts the acid is loosely held, 
for their solution strongly reddens litmus paper and their action 
is as if part of the acid were free in the salt. For instance, a 
solution of alum attacks iron giving ofi hydrogen, a soluble 
basic salt of aluminium being formed as well as sulphate of 
iron. The neutral salts of volatile acids give ofif acid simply by 
boiling their solutions, basic salts being formed. An aqueous 
solution of aluminium chloride loses its acid almost completely 
on evaporation. Gentle ignition is sufficient in most cases to 
completely decompose aluminium salts. Hydrated alumina 
dissolves easily in caustic alkali, forming soluble aluminates ; 
with baryta two aluminates are known, one soluble, the other 
not ; all other known aluminates are insoluble. 

Chemical reactions. — Neutral solutions of aluminium salts 
react as follows with the common reagents : 

Hydrogen sulphide produces no precipitate. 

Ammonium sulphide precipitates aluminium hydrate with 
separation of free sulphur. 

Caustic potash or soda precipitates aluminium hydrate, solu- 
ble in excess. 

Aqua ammonia precipitates aluminium hydrate insoluble in 
excess, especially in presence of ammoniacal salts. 

Alkaline carbonates precipitate aluminium hydrate insoluble 
in excess. 

Sodium phosphate precipitates white gelatinous aluminium 
phosphate, easily soluble in acids or alkalies. 

Before the blowpipe most aluminium compounds are infusible 
or are quickly converted into infusible alumina. In the light- 
colored, infusible compounds, the aluminium may be recognized 
by giving a beautiful, enamel-like blue color (Thenard's blue) 
when moistened with dilute solution of cobalt nitrate and 
ignited in a pure oxidizing flame. Light-colored fusible com- 
pounds will also give the same reaction, but in them the ab- 
sence of silica, boric and phosphoric acids must be proven by 
other tests before the presence of aluminium can be asserted 
with certainty. 

i i 2 aluminium. 

Aluminium Oxide. 

Commonly called alumina. Composition Al^Oj, and contains 
52.95 per cent, of aluminium when perfectly pure. Colorless 
corundum is a natural pure alumina, in which state it is infusible 
at ordinary furnace heats, insoluble in acids, has a specific 
gravity of 4, and is almost as hard as the diamond. To get 
this into solution it must be first fused with potassium hydrate 
or bisulphate. The alumina made by igniting aluminium hy- 
drate or sulphate is a white powder, easily soluble in acids if 
the ignition has been gentle, but becoming almost insoluble if 
the heat has been raised to whiteness. The specific gravity of 
this ignited alumina also varies with the temperature to which 
it has been raised; if simply to red heat, it is 3.75 ; if to bright 
redness, 3.8; and if to whiteness, 3.9. In the last case it ac- 
quires almost .the hardness of corundum. It can be melted to a 
clear, limpid liquid in the oxyhydrogen blow-pipe ; after cool- 
ing it forms a clear glass, often crystallized. Small artificial 
crystals of alumina, exactly similar to the natural ones, have 
been obtained by dissolving powdered alumina in fused boric 
oxide and then volatilizing the latter by an intense heat. 
Fluorine gas acts energetically on powdered alumina. As 
soon as it comes in contact, the whole mass becomes incandes- 
cent, forming a fluoride and disengaging oxygen. The reaction 

A\A + 6F = 2AIF3 + 30. 

disengages i,ioo calories of heat per kilogramme of fluorine 
acting. Gaseous chlorine does not act on it even at redness, 
but if carbon is present at the same time aluminium chloride is 
formed. Similarly, although neither carbon nor sulphur, alone 
or mixed together, acts on alumina, carbon bisulphide converts 
it into aluminium sulphide. 

The preparation of alumina for commercial purposes is de- 
scribed at length in the next chapter. 

properties of aluminium compounds. ii 3 

Aluminium Hydrates. 

There are three natural hydrates of aluminium, which may- 
be briefly described as follows : 

Diaspore, formula AI2O3.H2O or Al202.(OH)j, containing 85 
per cent, of alumina, occurs in crystalline masses as hard as 
quartz, with a specific gravity of 3.4. Bauxite, approximately 
of the formula AI2O3.2H2O or Al20.(OH)4, with the aluminium 
replaced by variable quantities of iron. If perfectly pure, it 
would contain 74 per cent, of alumina. Hydrochloric acid re- 
removes from it only the iron ; heated with moderately dilute 
sulphuric acid it gives up its alumina; a concentrated alkaline 
solution also dissolves the alumina. Calcined with sodium 
carbonate it forms sodium aluminate without melting. Gibbsite, 
formula AI2O3.3H2O or Al2(OH)6, containing, when pure sixty- 
five per cent, of alumina, is a mineral generally stalactitic, 
white, and with a specific gravity of 2.4. It loses two-thirds of 
its water at 300° and the rest at redness. 

The artificial hydrates are of two kinds, the soluble and in- 
soluble modifications. The latter is the common hydrate, such 
as is obtained by adding ammonia to a solution containing alu- 
minium. The precipitate is pure white, very voluminous, and 
can be washed free from the salts with which it was precipi- 
tated only with great difficulty. Its composition is Al2(OH)6, 
corresponding to the mineral gibbsite. It is insoluble in water, 
but easily soluble in dilute acids or alkali solutions. It dis- 
solves in small quantity in ammonia, but the presence of am- 
monia salts counteracts this action. When dissolved in caustic 
potash or soda the addition of ammoniacal salts reprecipitates it. 
It loses its water on heating, in the same manner as gibbsite. 
Many other properties of this hydrate, and its manufacture on 
a large scale, are given in the next chapter. The soluble modi- 
fication can only be made by complicated processes, too long 
to be described here, and is principally of use in the dyeing 
industries ; a full description can be found in any good chemi- 
cal dictionary. 

114 aluminium. 


Potassium aluminate. — Formula KjAljO^, crystallizes with 
three molecules of water, the crystals containing forty per cent, 
alumina, 37.5 per cent, potassa and 21.5 per cent, of water. It 
is formed when precipitated alumina is dissolved in caustic 
potash, or by melting together alumina and caustic potash in a 
silver dish and dissolving in water. If the solution is evapo- 
rated in vacuo, brilliant hard crystals separate out. They are 
soluble in water but insoluble in alcohol. 

Sodium aluminate has not been obtained crystallized. Ob- 
tained in solution by dissolving alumina in caustic soda or by 
fusing alumina with caustic soda or sodium carbonate and dis- 
solving in water. If single equivalents of carbonate of soda 
and alumina are used, the aluminate seems to have the com- 
position NajAljOi) if an excess of soda is used, the solution 
appears to contain Al2(0Na)e, or AljOj.sNajO. Dr. Bayer 
thinks that Al203.2Na20 and Al203.6Na20 are also possible in 
solution. If a solution of sodium aluminate is concentrated to 
20° or 30° B., alumina separates out; if carbonic acid gas is 
passed through it, aluminium hydrate is precipitated. For a 
description of its' manufacture on a large scale, see next 

Barium aluminate. — Formula BaAl^Oi. Deville prepared it 
by calcining a mixture of nitrate or carbonate of barium with • 
an excess of alumina, or by precipitating sulphate of aluminium 
in solution by baryta water in excess. The aluminate is solu- 
ble in about ten times its weight of water and crystallizes out 
on addition of alcohol. The crystals contain four molecules of 
water. Gaudin obtained it by passing steam over a mixture of 
alumina and barium chloride, or of alumina, barium sulphate, 
and carbon, at a red heat. Tedesco claimed that by heating 
to redness a mixture of alumina, barium sulphate, and carbon, 
barium aluminate was extracted from the residue by washing 
with water. He utilized this reaction further by adding solu- 
tion of alkaline sulphate, barium sulphate being precipitated 


(which was used over), while alkaline aluminate remained in 

Calcium, aluminate. — Lime water precipitates completely a 
solution of potassium or sodium aluminate, insoluble gelatinous 
calcium aluminate being formed, of the formula A]2(06Ca3) or' 
AljOs.sCaO. At a red heat it melts to a glass, which, treated 
after cooling with boiling solution of boric acid, affords a com- 
pound appearing to contain aAljOj-SCaO. (Tissier.) Lime 
water is also completely precipitated by hydrated alumina, the-^^ 
compound formed having the composition CaAljOi or AI2O3.-- 
CaO. Also, by igniting at a high temperature an intimate 
mixture of equal parts of alumina and chalk, Deville obtained^ 
a fused compound corresponding to the formfila CaAljO^. 

Zinc aluminate occurs in nature as the mineral Gahnit^, 
formula ZnALOi. Berzelius has remarked that when a solution 
of zinc oxide in ammonia and a saturated solution of alumina 
in caustic potash are mixed, a compound of the two oxides is 
precipitated, which is redissolved by an excess of either alkali. 

Copper aluminate. — On precipitating a dilute solution of 
sodium aluminate with an ammoniacal solution of copper sul- 
phate, the clear solution remaining contained neither copper 
nor aluminium. Whether the precipitate contained these com- 
bined as an aluminate I did not determine. 

Magnesium aluminate occurs in nature as Spinel; iron 
aluminate as Hercynite; beryllium aluminate as Chrysoberyl. 
The first mentioned has been lately proven to exist in small 
quantities in crystallized blast-furnace slags.* Ebelman has 
also prepared it in small crystals by dissolving alumina and 
magnesia in fused boric oxide and driving off some of the 
solvent by intense heat. 

Aluminium Chloride. 

Formula AICI3, contains 20.2 per cent, of aluminium. The 
commercial chloride is often yellow or even red from the pres- 
ence of iron, but the pure salt is quite white. It absorbs water 

* P. W. Shimer. Journal Am. Chemical Soc, 1894, p. 501. 


very rapidly from the air. It usually sublimes without melting, 
■especially when in small quantity, but if a large mass is rapidly 
ieated, it may melt and even boil, but its melting point is very 
close to its boiling point. Friedel and Crafts have determined 
the melting point as 178°, boiling point 183°. When sublimed 
it deposits in brilliant, hexagonal crystals. A current of steam 
rapidly decomposes it into alumina and hydrochloric acid. 
Oxygen disengages chlorine from it at redness, but decomposes 
it incompletely. Potassium or sodium decomposes it ex- 
plosively, the action commencing below redness. Anhydrous 
sulphuric acid converts it into aluminium sulphate. Alumin- 
ium chloride combines with many other chlorides, forming the 
double salts. 

On dissolving this salt in water, or by dissolving alumina in 
hydrochloric acid, a solution is obtained which on evaporation 
deposits crystals having the formula AICI3.6H2O. If these 
crystals are heated, they decompose, losing both water and 
acid and leaving alumina. Thus it is not possible to obtain 
anhydrous aluminium chloride by evaporating its solution, and 
the anhydrous salt must be made by other methods, detailed at 
length in the next chapter. 

Aluminium-Sodium Chloride. 

formula AlClj.NaCl contains 14 per cent, of aluminium. 
The commercial salt is often yellow or brown from the presence 
,of ferric chloride, but the pure salt is perfectly white. Its melt- 
ing point has been generally stated to be 180°, but Mr. Baker, 
<chemis.t for the Aluminium Company of London, states that 
-when .the absolutely pure salt is warmed it melts at 125° to 
130°. Xhat chemists should for thirty years have made an 
.error of this magnitude seems almost incredible, and it would 
be satisfactory if Mr. Baker would advance some further in- 
formation than the bare statement above. This salt volatilizes 
at a red heat without decomposition. It is less deliquescent in 
the air .than aluminium chloride, and for this reason is much 
icagier .to handle on a large scale. It is recently stated that the 


absolutely pure salt deteriorates less than the impure salt in the 
air, and the inference is drawn that perhaps the greater deli- 
quescence of the impure salt is due to the iron chlorides pres- 
ent. Its solution in water behaves similarly to that of alumin- 
ium chloride ; it cannot be evaporated to dryness without 
decomposition, the residue consisting of alumina and sodiun:i 

The manufacture of this double salt on a large scale is de- 
scribed in the next chapter. It may be prepared in the labora- 
tory by melting a mixture of the two component salts in the 
proper proportions. A similar salt with potassium chloride 
may be prepared by exactly analogous reactions. 

Aluminium-Phosphorus Chloride. 

Formula 2AICI3.PCI5, contains 9 per cent, of aluminium. It 
is a white salt, easily fusible, volatilizes only about 400° and 
sublimes slowly, fumes in the air and is decomposed by water. 
Produced by heating the two chlorides together or by passing 
vapor of phosphorus perchloride over alumina heated to redness. 

Aluminium-Sulphur Chloride. 

Formula 2AICI3.SCI4, contains 12.2 per cent, of aluminium. 
It forms a yellow crystalline mass, fuses at 100°, may be dis- 
tilled without change, and is decomposed by water. May be 
obtained by distilling a mixture of aluminium chloride and 
ordinary sulphur chloride, SClj. 

Aluminium-Selenium Chloride. 

Formula 2AlCl3.SeCl4. Obtained by heating the separate 
chlorides together in a sealed tube, when on careful distillation 
the less volatile double chloride remains. It is a yellow mass, 
melting at 100° and decomposed by water. 

Aluminium-Ammonium Chloride. 

Formula 2AICI3.3NH3. Solid aluminium chloride absorbs 
ammonia in large quantity, the heat developed liquefying the 


resulting compound. It may be sublimed in a current of 
hydrogen, but loses ammonia thereby and becomes 2AICI3.NH3. 


Formed by subliming aluminium chloride in a current of 
hydrogen sulphide. A current of hydrogen removes the ex- 
cess of the gas used, leaving on sublimation fine colorless 
crystals. In air it deliquesces rapidly and loses hydrogen 


Apparently of the formula 6AICI3.PH3. If phosphuretted 
hydrogen is passed over cold aluminium chloride very little is 
absorbed, but at its subliming point it absorbs a large quantity, 
the combination subliming and depositing in crystals. It is 
decomposed by water or ammonium hydrate, disengaging 
hydrogen phosphide. 

Aluminium Bromide. 

Formula AlBrg, containing lo.i per cent, of aluminium. It 
is colorless, crystalline, melts at 93° to a clear fluid which boils 
at 260°. It is still more deliquescent than aluminium chloride. 
At a red heat in contact with dry oxygen, it evolves bromine 
and forms alumina ; it is also decomposed slowly by the oxy- 
gen of the air. It dissolves easily in carbon bi-sulphide, the 
solution fuming strongly in the air. It reacts violently with 
water, the solution on evaporation depositing the compound 
AlBr3.6H20. The same result is attained by dissolving alumina 
in hydrobromic acid and evaporating. This hydrated chloride 
is decomposed by heat, leaving alumina. The specific gravity 
of solid aluminium bromide is 2.5. 

This compound is obtained by heating aluminium and bro- 
mine together to redness, or by passing bromine vapor over a 
mixture of alumina and carbon at bright redness. 

Aluminium Iodide. 
Formula AII3, containing 6.6 per cent, of aluminium. This 


compound is a white solid, fusible at 125°, and boils at 350°. 
It dissolves easily in carbon bisulphide, the warm saturated 
solution depositing it in crystals on cooling. It dissolves also 
in alcohol and ether. Its behavior towards water is exactly 
analogous to that of aluminium bromide. It is prepared by 
heating iodine and aluminium together, with special precau- 
tions to avoid explosion. 

Aluminium Fluoride. 

Formula AIF3, containing 32.7 per cent, of aluminium. It 
is sometimes obtained in crystals which are colorless and 
slightly phosphorescent. They are insoluble in acids, even in 
boiling sulphuric, and boiling solution of potash scarcely at- 
tacks them ; they can only be decomposed by fusion with 
sodium carbonate at a bright red heat. Melted with boric 
acid, aluminium fluoride forms crystals of aluminium borate. 
L. Grabau describes the aluminium fluoride which he obtains 
in his process as being a white powder, unalterable in air, 
unaffected by keeping, insoluble in water, infusible at redness, 
but volatilizing at a higher temperature. 

Deville first produced this compound by acting on alumin- 
ium with silicon fluoride at a red heat. He afterwards obtained 
it by moistening pure calcined alumina with hydrofluoric 
acid, drying and introducing into a tube made of gas carbon, 
protected by a refractory envelope. The tube was heated to 
bright redness, a current of hydrogen passing through mean- 
while to facilitate the volatilization of the fluoride. Brunner 
demonstrated that aluminium fluoride is formed and volatilized 
when hydrofluoric acid gas is passed over red-hot alumina. 
Finally, if a mixture of fluorspar and alumina is placed in car- 
bon boats, put into a carbon tube, suitably protected, heated to 
whiteness and gaseous hydrofluoric acid passed over it, alu- 
minium fluoride will volatilize and condense in the cooler part 
of the tube in fine cubical crystals, while calcium fluoride re- 
mains in the boats. 

i20 aluminium. 

Aluminium Fluorhydrate. 
When calcined alumina or kaolin is treated with hydrofluoric 
acid, alumina being in excess, soluble fluorhydrate of alumin- 
ium is formed, which deposits on evaporating the solution. It 
has the formula 2AIF3.5H2O, and easily loses its water when 

Aluminium-Hydrogen Fluoride. 

If to a strongly acid solution of alumina in hydrofluoric acid 
alcohol is added, an oily material separates out and crystallizes, 
having the formula 3AIF3.2HF.5H2O. If the acid solution is 
simply evaporated, acid fumes escape and a crystalline mass 
remains which, washed with boiling water and dried, has the 
formula 4AIF3.HF.10H2O. On heating these compounds to 
400° or 500° in a current of hydrogen, pure amorphous alu- 
minium fluoride remains. The acid solution of alumina first 
used seems to contain an acid of the composition AIF3.3HF, 
which is capable of forming salts with other bases. Thus, if 
this solution is neutralized with a solution of soda, a precipitate 
of artificial cryolite, AlFs.sNaF, falls. The similar potash 
compound is formed in the same way. 

Aluminium-Sodium Fluoride. 

Formula AlFs.sNaF, containing 12.85 P^r cent, of aluminium, 
occurs native as cryolite, a white mineral with a waxy appear- 
ance, as hard as calcite, specific gravity 2.9, melting below 
redness and on cooling looking like opaque, milky glass. If 
kept melted in moist air, or in a current of steam, it loses 
hydrofluoric acid and sodium fluoride and leaves a residue of 
pure alumina. When melted it is decomposable by an electric 
current or by sodium or magnesium. It is insoluble in water, 
unattacked by hydrochloric but decomposed by hot sulphuric 
acid. The native mineral is contaminated with ferrous carbon- 
ate, silica, phosphoric and vanadic acids. An extended de- 
scription of its utilization, manufacture, etc., will be found in 
the next chapter. 


A native compound having the formula sAlFj.jNaF is snow- 
white, very similar to cryolite in appearance and properties, 
except that it crystallizes in a different form and is more 
fusible. It is called Chiolite. The mineral Pachnolite is also 
similar, except that part of the sodium is displaced by calcium. 

Aluminium Sulphide. 

Formula AI2S3, containing 36 per cent, of aluminium. The 
pure salt is light yellow in color and melts at a high tempera- 
ture. In damp air it swells up and disengages hydrogen sul- 
phide, forming a grayish white powder ; it decomposes water 
very actively, forming hydrogen sulphide and ordinary gela- 
tinous aluminium hydrate. Steam decomposes it easily, at red 
heat forming amorphous alumina, which is translucent and very 
hard. Gaseous hydrochloric acid transforms it into aluminium 
chloride. Elements having a strong affinity for sulphur reduce 
it, setting free aluminium, but it is doubtful if hydrogen or 
carburetted hydrogen has this efifect. 

It may be formed by throwing sulphur into red-hot alumin- 
ium, or by passing sulphur vapor over red-hot aluminium. 
Traces only of aluminium sulphide are formed by passing 
hydrogen sulphide over ignited alumina, but carbon-bisulphide 
vapor readily produces this reaction. For details of its forma- 
tion see next chapter. 

Double sulphides of aluminium and sodium or potassium,, 
similar in composition to the double chlorides, may be made 
by melting the two sulphides together, or, more easily, by add- 
ing sodium or potassium carbonate to the ignited alumina 
which is in process of conversion into sulphide. These salts 
are easily fusible and nearly unchangeable in the air, but a 
more detailed description of them is wanting. They are said 
to be quite suitable for electrolytic decomposition. (See 
Bucherer's electrolytic process.) 

Aluminium sulphide has recently come into use in chemical 
laboratories as a very convenient means of generating hydrogen 
sulphide. A London chemical firm manufactures it by stirring 


aluminium into molten sulphide of lead ; the lead separating 
out sinks to the bottom, while the aluminium sulphide floats. 
The sulphide is sold at $0.6o per pound, and is said to gen- 
erate the gas as cheaply as by the ordinary methods. 

Aluminium Selenide. 

When aluminium is heated in selenium vapor, the two ele- 
ments combine with incandescence, producing a black powder. 
In the air this powder evolves the odor of hydrogen selenide ; 
in contact with water it disengages that gas abundantly, and 
furnishes a red deposit of selenium along with aluminium 
hydrate. When a solution of an aluminium salt is treated with 
an alkaline poly-selenide, a flesh-colored precipitate falls, the 
composition of which is not known, which is decomposed at 
redness, leaving aluminium. 

Aluminium Borides. 

AIB2, containing 55.1 per cent, of aluminium, was first ob- 
tained by Deville and Wohler by heating boron in contact with 
aluminium, or by reducing boric acid with the latter metal, the 
action not being long continued. Also, if a current of boron 
trichloride with carbonic oxide is passed over aluminium in 
boats in a tube heated to redness, aluminium chloride volatil- 
izes and there remains in the boats a crystalline mass, cleav- 
able, and covered with large hexagonal plates of a high metal- ' 
lie lustre. To remove the aluminium present in excess, the 
mass is treated with hydrochloric acid and then with caustic 
soda. The final residue is composed of hexagonal tablets, very 
thin but perfectly opaque, of about the color of copper. These 
crystals do not burn in the air, even if heated to redness, but 
their color changes to dark-gray. They burn in a current of 
chlorine, giving chlorides of the two elements contained in 
them. They dissolve slowly in concentrated hydrochloric acid 
or in solution of caustic soda ; nitric acid, moderately concen- 
trated, attacks them quickly. 

AlBs, containing 45 per cent, of aluminium, has been ob- 


tained by Hampe by heating aluminium with boric acid for 
three hours at a high temperature, carbon being carefully kept 
away. On cooling very slowly, the upper part of the fusion is 
composed of aluminium borate, the centre is of very hard 
alumina containing a few black crystals of aluminium boride, 
while at the bottom is a button of aluminium also containing 
these crystals. To free these crystals, the aluminium is dis- 
solved by hydrochloric acid. These crystals are the compound 
sought for, and contain no other impurity than a little alumina, 
which can be removed by boihng sulphuric acid. These puri- 
fied crystals are black, but are thin enough to show a dark-red 
by transmitted light. Their specific gravity is 2.5, they are 
harder than corundum, but are scratched by the diamond. 
Oxygen has no action on them at a high temperature, solution 
of caustic potash or hydrochloric acid does not attack them, 
boiling sulphuric acid has scarcely any action, but they dissolve 
completely in warm, concentrated nitric acid. 

If the operation by which this product is made is conducted 
in the presence of carbon, the compound formed contains less 
aluminium and also some carbon. Its composition corresponds 
to the formula Al3C2B«, containing about 12 per cent, of alu- 
minium and 3.75 per cent, of carbon. The crystals of this 
compound are yellow and as brilliant as the diamond. Their 
specific gravity is 2.6, hardness between that of corundum and 
the diamond. They are not attacked by oxygen, even at a 
high temperature; hot hydrochloric or sulphuric acid attacks 
them only superficially, concentrated nitric acid dissolves them 
slowly but completely. They resist boiling solution of caustic 
potash or fused nitre, but take fire in fused caustic potash or 
chromate of lead. 

Aluminium and Carbon. 

Deviile stated that he was unable to combine carbon with 
aluminium. On decomposing carbon tetrachloride by aluniin- 
ium, ordinary carbon was formed, while the aluminium remain- 
ing was unchanged. The Cowles Company obtain in their 


electric furnace, when reducing a mixture of alumina and car- 
bon alone, a yellow, crystalline substance which was exhibited 
by Dr. T. Sterry Hunt,* as an alloy of aluminium and carbon,, 
but that this is the case is not yet accepted as certain. 

On dissolving impure aluminium in solution of caustic pot- 
ash, a black residue was obtained which behaved, when filtered' 
out and dried, partly like amorphous carbon. I have heard 
it stated that molten aluminium does not dissolve appreciable- 
quantities of carbon, and that its properties are affected con- 
siderably thereby; but I am not able to give any further light 
on the subject than the above experiment, which seemed to- 
show considerable carbon in an impure metal. This is a sub- 
ject which needs thorough investigation. 

Since the observation just given was made, in 1890, the fact 
that this residue does contain amorphous carbon has been es- 
tablished by Le Verrier (see p. 57), who has also confirmed 
Hunt's statements as to an aluminium carbide. Le Verrier 
states I that up to his own experiments this compound was 
unknown, but he must have been ignorant of Hunt's observa- 
tions. The method used by Le Verrier was as follows: Little 
boats of carbon were filled with aluminium, put into a carbon 
tube through which a current of hydrogen was passed, and 
heated in the electric furnace. After cooling, the aluminium 
was of a grey color, and on breaking it was seen to be strewn 
with brilliant fine-yellow crystals. To separate these out, the 
aluminium is dissolved by concentrated hydrochloric acid, sur- 
rounding the vessel with ice to keep the temperature low. The 
residue is washed with ice-cold water. This whole operation 
must be conducted as quickly as possible, since the carbide is 
attacked by water. 

The carbide thus prepared has a specific gravity, taken in 
benzine, of 2.36. It is decomposed in the highest heat of an 
electric furnace, attacked by chlorine at a red-heat, but only 

♦Halifax Meeting, Am. Ins. Mining Engineers, Sept. i6, 1885. 
tComptes Rendus, July, 1894, p. 16. 


superficially by oxygen, since the layer of alumina formed pro- 
tects the part underneath. Violently attacked by sulphur at 
that temperature, also by permanganate of potassium, bichro- 
mate of potassium, chromic acid and oxide of lead ; potassium 
chlorate has no action. Caustic potash in fusion attacks it very 
energetically at about 300°. It slowly decomposes water in 
the cold with a very curious result, namely, the formation of 
jnethane gas, according to the reaction : 

QAl^ + 12H2O = aCH, -f- 4A1(0H)3. 

An analysis of these crystals gave the following result : 

Aluminium. Carbon. 

7448 23.5 

7.5.12 24.2 

74-7 24.7 

74.9 24.8 


(Required by the formula AI4C3.) 

75.4 24.6 

Aluminium Nitride. 

Formula AIN, containing 66 per cent, of aluminium, is 
iormed when aluminium is heated in a carbon crucible to a 
high temperature. Mallet* obtained it unexpectedly when 
heating aluminium with dry sodium carbonate at a high heat, 
lor several hours, in a carbon crucible. The aluminium is par- 
tially transformed into alumina, some sodium vaporizes, and 
some carbon is deposited. After cooling, there are found on 
the surface of the button little yellow crystals and amorphous 
drops, to recover which the whole is treated with very dilute 
hydrochloric acid. This product has the composition AIN. 
"Calcined in the air it slowly loses nitrogen and forms alumina. 
It decomposes in moist air, loses its transparency, becomes a 
lighter yellow, and finally only alumina remains, the nitrogen 

*Ann. der Chemie u. Pharmacie, 186, p. 155. 


having formed ammonia. Melted with caustic potash it dis- 
engages ammonia and forms potassium aluminate. 

Aluminium Sulphate. 

Anhydrous. — The salt obtained by drying hydrated alumin- 
ium sulphate at a gentle heat has the formula Al2(S04)3, con- 
taining 15.8 per cent, of aluminium, and of a specific gravity 
of 2.67. By heating this salt several minutes over a Bunsen 
burner it loses almost all its acid, leaving alumina. Hydrogen 
likewise decomposes it at redness, forming water and sulphur 
dioxide and leaving alumina with hardly a trace of acid. 
Melted with sulphur, Violi states that it is transformed into 
aluminium sulphide, evolving sulphurous acid gas.* Hot 
hydrochloric acid in excess partly converts it into aluminium 

Hydrated. — This is the ordinary aluminium sulphate; its 
formula is Al2(S04)3.i6H20, and it contains 8.0 per cent, of 
aluminium, equal to 14.5 per cent, of alumina, and 47 per cent, 
of water. The pure salt is not hygroscopic. In the presence 
of impurities the amount of water increases to 18 H2O and the 
salt becomes hygroscopic. It has a white, crystalline appear- 
ance, and tastes like alum. It dissolves freely in water, from 
which it crystallizes out at ordinary temperatures with the 
above formula ; crystallized out at a low temperature it retains 
27H2O, or one-half as much again. Water dissolves one-half 
its weight of this salt, the solution reacting strongly acid ; it is 
almost insoluble in alcohol. At a gentle heat it melts in its 
water of crystallization, then puffs up and leaves a porous mass 
of anhydrous sulphate which is soluble with difficulty in water. 
If heated to redness it leaves only alumina. The salt with 
16H2O has a specific gravity of 1.76; it is the salt found in 
fibrous masses in solfataras, its mineralogical name being Halo- 
trichite. A hydrated sulphate with 10H2O is formed and pre- 
cipitated when alcohol is added to an aqueous solution of alu- 

* Berichte der Deutschen Chemischen Gesellschaft, X, 293. 


minium sulphate. On heating it acts similarly to the other 
hydrated sulphates. 

Basic. — On precipitating a solution of aluminium sulphate 
with alkaline hydrate or carbonate, a series of basic salts are 
formed. On precipitating with ammonia, the compound formed 
has the formula AljO3.SO3.9HjO, corresponding to the mineral 
Aluminite. If the ammonia is in insufficient quantity to en- 
tirely precipitate the solution, a precipitate is very slowly 
formed having the formula 3AI2O3.2SO3.20H2O. On precipitat- 
ing a cold solution of alum by alkaline carbonate not in excess, 
a precipitate is very slowly formed having the formula 2A1203.- 
SO3.12H2O. If a very dilute solution of acetate of alumina is 
precipitated by adding potassium sulphate, a compound de- 
posits very slowly having the formula 2AI2O3.SO3.10H2O. Na- 
tive minerals are met with of analogous composition to these 
precipitates: Felsobanyte, 2AI2O3.SO3.10H2O ; Paraluminite, 
2AI2O3.SO3.ISH2O. By heating a concentrated solution of alu- 
minium sulphate with aluminium hydrate, and filtering cold, 
the solution deposits on further cooling a gummy mass having 
the formula Al203.2S03.;irH20. On washing with water it de- 
posits a basic salt having the formula AI2O3.SO3. By letting 
stand a very dilute solution of sulphuric acid completely satu- 
rated with aluminium hydrate, Rammelsberg obtained trans- 
parent crystals having the formula 3AI2O3.4SO3.30H2O. On 
boiling a solution of aluminium sulphate with zinc, Debray 
obtained a granular precipitate having the formula 5AI2O3.3SO3.- 
20H2O. By leaving zinc a long time in a cold solution of alu- 
minium sulphate, a gelatinous precipitate was obtained having 
the formula 4AI2O3.3SO3.36H2O ; the same compound was 
formed if the zinc was replaced by calcium carbonate. 

The manufacture of aluminium sulphate from clay, aluminous 
earths, cryolite, bauxite, etc. , is carried on industrially on a 
very large scale ; descriptions of the processes used may be 
found in any work on industrial chemistry — they are too foreign 
to metallurgical purposes to be treated of here. 

128 aluminium. 


Under this name are included a number of double salts con- 
taining water, crystallizing in octahedra and having the general 
formula R2S04.R2(S04)3.24H20, in which the first R may be 
potassium, sodium, rubidium, caesum, ammonium, thallium or 
even organic radicals ; the second R may be aluminium, iron, 
manganese or chromium ; the acid may even be selenic, chro- 
mic or manganic, instead of sulphuric. We will briefly describe 
the most important alums consisting of double sulphates of 
aluminium and another metal, remarking, as with aluminium 
sulphate, that their preparation may be found at length in any 
chemical treatise. 

Potash Alum. 

Formula K2S04.Al2(S04)3.24H20, containing 10.7 per cent, of 
alumina or 5.7 per cent, of aluminium. Dissolves in 25 parts 
of water at 0° and in two-sevenths part at 100°. The solution 
reacts acid. It forms colorless, transparent octahedrons, insol- 
uble in alcohol. On exposure to air they become opaque, 
being covered with a white coating, which is said not to be 
efflorescence — a loss of water — but to be caused by absorption 
of ammonia from the air. The crystals melt in their water of 
<;rystallization, but lose it all above 100°. Heated to redness 
it swells up strongly, becomes porous and friable, giving the 
product called calcined alum; at whiteness it loses a large part 
of its sulphuric acid, leaving a residue of potassium sulphate and 
alumina. If it is mixed with one-third its weight of carbon and 
calcined, the residue inflames spontaneously in the air. If a 
mixture of alumina and bisulphate of potassium is fused and 
afterwards washed with warm water, a residue is obtained of 
anhydrous alum, or K2S04.Al2(S04)3. The mineral Alunite is 
a basic potash alum, K2S04.3(A1A-S0,).6H20. 

Ammonia Alum. 

Formula (NH0,SO4.Al2(SOl)s.24HjO, containing 11.3 per 
■cent, of alumina or 6.0 per cent, of aluminium. Dissolves in 


20 parts of water at 0° and in one-fourth part at 100°. When 
heated, the crystals swell up strongly, forming a porous mass, 
losing at the same time water and sulphurous acid ; if the tem- 
perature is high enough there remains a residue of pure alu- 
mina. The temperature necessary for complete decomposition 
is higher than that required for volatilizing ammonium sulphate 

Soda Alum. 

Formula Na2S04.Al2(S04)3.24H20, containing 11. i per cent, 
of alumina or 5.9 per cent, of aluminium. Dissolves in an 
equal weight of water at ordinary temperatures. The crystals 
effloresce and fall to powder in the air. It is insoluble in abso- 
lute alcohol. On account of its great solubility in water it 
cannot be separated from ferrous sulphate by crystallization, and 
therefore it is either contaminated with much iron or else, to 
be obtained pure, special expensive methods must be adopted. 
These diflSculties cause the manufacture of soda alum to be in- 
significant in amount when compared with potash alum. 

Aluminium-Metallic Sulphates. 

Sulphate of aluminium forms double sulphates with iron, 
manganese, magnesium and zinc, but these compounds are not 
analogous to the alums. They are extremely soluble in water, 
do not crystallize in octahedrons or any isometric forms, and 
their composition is different from the alums in the amount of 
water of crystallization. It has heen determined that the double 
sulphates with manganese and zinc contain 25 equivalents of 
water, which would permit their being considered as combina- 
tions of sulphate of aluminium, Al2(S04)3.i8H20, with a sul- 
phate of the magnesian series containing seven equivalents of 
water, as ZnS04.7H20. 

Aluminium Selenites. 

By adding a solution of selenite of soda to one of sulphaite 
of aluminium maintained in excess, an amorphous, voluminous 


precipitate forms, having the composition 4Al2O3.9SeO2.3H2O. 
This substance decomposes on being heated, leaving alumina. 
If varying quantities of selenious acid are added to this first 
salt, other salts of the formulas Al2O3.3SeO2.7H2O, 2Al203.9Se02. 
12H2O, Al2O3.6SeO2.SH2O, are formed. These are mostly in- 
soluble in water and decompose on being heated, like the first. 

Aluminium Nitrate. 

Formula A1(N03)3.9H20 is obtained on dissolving aluminium 
hydrate in nitric acid. If the solution is evaporated keeping it 
strongly acid, it deposits on cooling voluminous crystals having 
the formula 2Al(N03)3.isH20. This salt is deliquescent, melts 
at 73° and gives a colorless liquid which, on cooling, becomes 
crystalline. It is soluble in water, nitric acid, and alcohol ; on 
evaporating these solutions it is obtained as a sticky mass. It 
is easily decomposed by heat; if kept at 100° for a long time 
it loses half its weight, leaving as residue a soluble salt of the 
formula 2AI2O3.3N2O5.3H2O. Carried to 140°, this residue 
loses all its nitric acid, leaving alumina. On this property is 
based a separation of alumina from lime or magnesia, since the 
nitrates of these latter bases resist the action of heat much 
better than aluminium nitrate. 

Aluminium Antimonate. 

Al203.3Sb205.isH20 can be prepared by double decomposi- 
tion from potassium antimonate and a soluble aluminium salt. 
It deposits from solution after several days' standing in shin- 
ing, microscopic crystals. At 100° it loses sHjO, at 150°, 2^ 
H2O more, at 200° it retains only 3H2O and becomes incandes- 
cent. Al2O3.Sb2O5.9H2O has been also described. 

Aluminium Phosphates. 

The normal phosphate, AIPO4, is obtained as a white, gelat- 
inous precipitate when a neutral aluminium solution is treated 
with sodium phosphate. It is soluble in alkalies or mineral 
acids, but not in acetic acid. If a solution of this salt in acids is 


neutralized with ammonia, a basic phosphate is precipitated hav- 
ing the composition 3Al,(OH)3PO,-|-2Al(OH)3. The mineral 
Wavellite has this composition, with nine molecules of water. 
The mineral Kalait contains AlP04+Al(OH)3+H20, and when 
it is colored azure blue by a little copper it forms the Turquois. 
Aluminium meta-phosphate has the formula AIP3O9 or AI2O3.- 

Aluminium Carbonate. 

If to a cold solution of alum a cold solution of sodium carbon- 
ate is added drop by drop, stirring constantly until the solution 
reacts feebly alkaline, a precipitate is obtained which, after be- 
ing washed with cold water containing carbonic acid gas, con- 
tains when damp single equivalents of alumina and carbonic 
acid, ALjOa.COij. If the precautions indicated are not used, the 
precipitate contains a very small proportion of carbonic acid. 

Sestini * states that one litre of water saturated with carbonic 
acid gas at the ordinary atmospheric pressure will dissolve 10 
milligrammes of alumina, the solution becoming turbid on 
boiling or agitation. This reaction is probably the key to the 
understanding of the geological formation of native alumina 

When a solution of sodium bi-carbonate is added to one of 
sodium aluminate, a double carbonate of the alkali and alumin- 
ium is precipitated, which is easily soluble in acids. (Mende- 

Aluminium Borate. 

Formula 3AI2OS.BO3. Prepared by Ebelman by heating 
together alumina, oxide of cadmium, and boric acid. After 
three days' heating the platinum capsule containing the mix- 
ture was found covered with transparent crystals of the above 
composition, hard enough to scratch quartz and having a spe- 
cific gravity of 3. Troost obtained the same substance by 
heating alumina in the vapor of boron trichloride. Fremy pre- 
pared it by heating fluoride of aluminium with boric acid. 

* Gazetta Chim., Vol. 20, p. 313. 


Ebelman also obtained it by heating a mixture of alumina and 
borax to whiteness ; under these conditions crystals of corun- 
dum were formed at the same time. 

By precipitating a cold solution of alum with sodium borate, 
double salts are obtained containing soda, but which leave on 
washing with warm water two compounds having the formulae 
zAlA-BOs.sH.O and 3Al203.2B03.8H,0. If the washing is 
prolonged too far the two salts are completely decomposed, 
leaving a residue of pure alumina. 

Aluminium Silicates. 

These compounds are very plentiful in nature, both hydrous 
and anhydrous. The addition of a soluble aluminium salt to a 
solution of sodium silicate would doubtless produce a precipi- 
tate of aluminium silicate. Many of 'the native silicates have 
been formed by the mingling of lime or alkaline water with 
cabonated water carrying alumina in solution. In the native 
silicates the silica and alumina occur in many proportions ; 
the following ratios of alumina to silica have been observed : 
2-1, 3-2, i-i, 2-3, 1-2, 1-3, 1-4, 1-8, thus varying from 
tri-basic silicates to pent-acid silicates. The pure silicates 
of aluminium alone are infusible before the blowpipe ; but 
the presence of other elements, especially the alkalies, calcium 
or iron, makes them quite easy to melt. Ordinary blast- 
furnace slags are silicates of aluminium, calcium and magne- 
sium, and vary from bi-acid to sesqui-basic silicates. In some 
of these more basic silicates part of the aluminium present may 
exert its acid character, and, in fact, the occurrence of crystals 
of magnesium aluminate has been noted in crystallized blast- 
furnace slag which was basic and rich in magnesium. 



We will consider this division under four heads : — 

I. Alumina. 
II. Aluminium chloride and aluminium-sodium chloride. 

III. Aluminium fluoride and aluminium-sodium fluoride. 

IV. Aluminium sulphide. 

The Preparation of Alumina. 

We will treat this subject in three divisions : — 

1 . From Aluminium Sulphate or Alums. 

2. From Bauxite. 

3. From CryoHte. 

I. Preparation of Alumina from Alums or Aluminium 


Hydrated alumina can be precipitated from a solution of any 
aluminium salt by ammonium hydrate, an excess of which re- 
dissolves a portion. Its chemical formula is ordinarily written 
AI2O3.3H2O or A^OHg). The aluminium hydrate thus pre- 
cipitated is a pure white, very voluminous, almost pasty mass, 
very hard to wash. By boiling and washing with boiUng water 
it becomes more dense, but always remain very voluminous. 
Washing on a filter with a suction apparatus gives the best re- 
sults. At a freezing temperature this hydrate changes into a 
dense powder which is more easily washed. On drying, it 
shrinks very much in volume and forms dense, white pieces, 
transparent on the edges. When dried at ordinary tempera- 
tures it has the composition AljOg.H^O. On ignition, the 



Other molecule of water is driven off, leaving anhydrous alu- 
mina. After gentle ignition it remains highly hygroscopic, and 
in a very short time will take up from the air 1 5 per cent, of 
water. In this condition it is easily soluble in hydrochloric or 
sulphuric acid. On stronger ignition it becomes harder and 
soluble only with difficulty in concentrated acid ; after ignition 
at a high temperature it is insoluble, and can only be brought 
into solution again by powdering finely and fusing with potas- 
sium acid-sulphate or alkaline carbonate. At ordinary furnace 
temperatures alumina does not melt, but in the oxy- hydrogen 
blow-pipe or the electric arc it fuses to a limpid Hquid and ap- 
pears crystalline on cooling. 

The precipitation in aqueous solution and subsequent ignition 
is not economical enough to be practised on a large scale, and 
for industrial purposes the aluminium sulphate or alum is ig- 
nited directly. About the easiest way to proceed is to take 
ammonia alum crystals, put them into a clean iron pan and 
heat gently, when the salt melts in its water of crystallization. 
When the water has evaporated, a brittle, shining, sticky mass 
remains, which on further heating swells up and decomposes 
into a dry, white powder. This is let cool, powdered, put into 
a crucible and heated to bright redness. All the ammonia and 
almost all the sulphuric acid are thus removed. The rest of 
the acid can be removed by moistening the mass with a solu- 
tion of sodium carbonate, drying and again igniting ; on wash- 
ing with water the acid is removed as sodium sulphate. The 
residue, however, will contain some caustic soda, which for its 
further use in making aluminium chloride is not harmful. Pot- 
ash alum can be treated in a similar way, the potassium sul- 
phate being washed away after the first ignition. Still more 
easily and cheaply can alumina be made by igniting a mixture 
of 4 parts aluminium sulphate and i of sodium carbonate. On 
washing, sodium sulphate is removed from the alumina.* 

Deville used the following method at Javel : Ammonia alum or 

* Kerl and Stohman, 4th Ed., p. 739. 


even the impure commercial aluminium sulphate was calcined, 
the residue appearing to be pure, white alumina, but it still con- 
tained sulphuric acid, potassium sulphate, and a notable pro- 
portion of iron. This alumina is very friable, and is passed 
through a fine sieve, and put into an iron pot with twice its 
weight of solution of caustic soda of 45 degrees. It is then 
boiled and evaporated, and the alumina dissolves even though 
it has been strongly calcined. The aluminate of soda produced 
is taken up in a large quantity of water, and if it does not show 
clear immediately, a little sulphuretted hydrogen is passed in, 
which hastens the precipitation of the iron. The liquor is let 
stand, the clear solution decanted ofT and subjected while still 
warm to the action of a stream of carbonic acid gas. This con- 
verts the soda into carbonate and precipitates the alumina in a 
particularly dense form which collects in a space not one- 
twentieth of the volume which would be taken up by gelatinous 
alumina. This precipitate is best washed by decantation, but a 
large number of washings are necessary to remove all the sodium 
carbonate from it ; it is even well, before finishing the washing, 
to add a little sal-ammoniac to the wash-water in order to hasten 
the removal of the soda. The well-dried alumina is calcined at 
a red heat. 

Tilghman* decomposes commercial sulphate of alumina, 
Al2(S04)3.i8H20, by filling a red-hot fire-clay cylinder with it. 
This cylinder is lined inside with a magnesia fettUng, is kept at 
a red heat, the sulphate put in in large lumps, and steam is 
passed through the retort, carrying with it vapor of sodium 
chloride. This last arrangement is effected by passing steam 
into a cast-iron retort in which the salt named is kept melted, 
and as the steam leaves this retort it carries vapor of the salt 
with it. It is preferable, however, to make a paste of the sul- 
phate of alumina and the sodium chloride, forming it into small 
hollow cylinders, which are well dried, and then the fire-clay 
cylinder filled with these. Then, the cylinder being heated to 

* Mierzinski. 


whiteness, highly superheated steam is passed over it. The 
hydrochloric acid gas which is formed is caught in a condensing 
apparatus, and there remains a mass of aluminate of soda, 
which is moistened with water and treated with a current of 
carbon dioxide and steam. By washing the mass, the soda 
goes fnto solution and hydrated alumina remains, which is 
washed well and is ready for use. 

Mr. Webster's process for making pure alumina at a low price 
was incorporated as a part of the Aluminium Co. Ld.'s pro- 
cesses. The only description we can give of it is dated 1883. 

* Three parts of potash alum are mixed with one part of 
pitch, placed in a calcining furnace and heated to 200° or 250°. 
About 40 per cent, of this water is thus driven off, leaving sul- 
phate of potash and aluminium, with some ferric oxide. After 
heating about three hours, the pasty mass is taken out, spread 
on a stone floor, and when cold broken to pieces. Hydrochloric 
acid (20 to 25 per cent.) is poured upon these pieces, placed 
in piles, which are turned over from time to time. When the 
evolution of sulphuretted hydrogen has stopped, about five per 
cent, of charcoal-powder or lampblack, with enough water to 
make a thick paste, is added. The mass is thoroughly broken 
up and mixed in a mill, and then worked into balls of about a 
pound each. These are bored through to facilitate drying, and 
heated in a drying chamber at first to 40°, then in a furnace 
from 95° to 150°. The balls are then kept for three hours at' 
a low red heat in retorts, while a mixture of two parts steam 
and one part air is passed through, so that the sulphur and car- 
bon are converted into sulphurous oxide and carbonic oxide, 
and thus escape. The current of gas carries over some potas- 
sium sulphate, ferrous sulphate and alumina, and is therefore 
passed through clay condensers. 

The residue in the retorts consists of alumina and potassium 
sulphate ; it is removed, ground to fine powder in a mill, 

♦Austrian Patent, Sept. 28, 1882; English patent. No. 2580, 1881. Dingier, 1883, 
vol. 259, p. 86. 


treated with about seven times its weight of water, boiled in a 
pan or boiler by means of steam for about one hour, then 
allowed to stand till cool. The solution containing the potas- 
sium sulphate is run ofifand evaporated to dryness, the alumina 
is washed and dried. The potassium sulphate, as a by-product, 
is said to pay one-half the cost of the process. 

This deposit contains about 84 per cent, of alumina, while 
that obtained by the old process of precipitation has only 65 
per cent. Thus a large saving is effected in cost, and 19 per 
cent, more alumina is obtained. In addition to this, the whole 
of the by-products are recovered, consisting of potassium sul- 
phate, sulphur (which is used in making sulphuric acid), and 
aluminate of iron. 

2. Preparation of Alumina from Bauxite. 

At Salindres, the alumina used in the Deville process was 
obtained from bauxite by the following processes, which are 
in general use for extracting pure alumina from this mineral. 
Bauxite is plentiful enough in the south of France, principally 
in the departments of Herault, Bouches-du-Rhone, and Var. 
It contains as much as seventy-five per cent, alumina. To 
separate the alumina from ferric oxide, it is treated with car- 
bonate of soda, under the influence of a sufiSciently high tem- 
perature, the alumina displacing the carbonic acid and forming 
an aluminate of soda, Al^Og.sNa^O, while the ferric oxide re- 
mains unattacked. A simple washing with water then permits 
the separation of the former from the insoluble ferric oxide. 
The bauxite is first finely pulverized by means of a vertical 
mill-stone, then intimately mixed with some sodium carbonate. 
The mixture is made, for one operation, of — 

480 kilos, bauxite. 

300 " sodium carbonate of 90 alkali degrees. 

This mixture is introduced into a reverberatory furnace, re- 
sembling in form a soda furnace, and which will bear heating 


Strongly. The mass is stirred from time to time, and it is kept 
heated until all the carbonate has been attacked, which is 
recognized by a test being taken which does not effervesce 
with acids. The operation lasts from five to six hours. 

The aluminate thus obtained is separated from ferric oxide by 
a washing with warm water. This washing is made at first with 
a feeble solution which has served for the complete exhaustion 
of the preceding charge, which was last washed with pure 
water, forming thus this feeble solution. This gives, on the 
first leaching, solutions of aluminate concentrated enough to be 
called strong liquor, which are next treated by the current of 
carbonic acid gas to precipitate the hydrated alumina. The 
charge is next washed with pure water, which completely re- 
moves the aluminate ; this solution is the weak liquor, which is 
put aside in a special tank, and used as the first leaching liquor 
on the next charge treated. This treatment takes place in the 
following apparatus (see Fig. i ) : ^ is a sheet-iron vessel, in 
the middle of which is a metallic grating, F, on which is held 
all round its edges, by pins, a cloth, serving as a filter. A 
ought to be closed by a metallic lid held on firmly by bolts. 
To work the apparatus, about 500 kilos of the charge to be 
washed is placed on the filter cloth, the lid is closed, then 
the steam-cock / of the reservoir A is opened. In A is the 
weak solution from the last washing of the preceding charge. 
The pressure of the steam makes it rise by the tube T into the 
filter; another jet of steam, admitted by the cock b, rapidly 
warms the feeble liquor as it soaks into the charge. After fil- 
tering through, the strong liquor is drawn off by turning the 
stop-cock G. The weak solution of the reservoir A is put into 
the filter in successive portions, and not all at once ; and after 
each addition of solution has filtered through, its strength in 
B.° is taken, before any more solution is run in ; then, when the 
solution marks 3° to 4°, it is placed in a special tank for weak 
liquor, with all that comes through afterwards. Just about this 
time, the weak liquor of the reservoir A is generally all used up, 
and is replaced by pure water introduced by the tube d. All 



the solutions which filtered through marking over 3° to 4° B., 
are put together, and form a strong liquor which marks about 
12° B. This extraction of the aluminate being completed by 
the pure water, the residue on the filter is taken out, and a new 
operation may be commenced. 

Fig. I. 




Li— p 












i-cvrrrpe O 

The strong liquor is introduced into a vessel having an agi- 
tator, where a strong current of carbonic acid gas may precipi- 
tate the alumina from it. The gas is produced by small streams 
of hydrochloric acid continuously falling on some limestone 
contained in a series of earthenware jars. The precipitation 
vessel is called a baratte. The carbonic acid after having 
passed through a washing-flask, is directed to a battery of three 
barattes, where the precipitation is worked methodically, so as 
to precipitate completely the alumina of each baratte, and util- 



ize at the same time all the carbon dioxide produced. In order 
to do this, the gas always enters first into a baratte in which 
the precipitation is nearest completion, and arrives at last to 
that in which the solution is freshest. When the gas is not all 
absorbed in the last baratte, the first is emptied, for the precip- 
itation in it is then completed, and it is made the last of the 
series, the current being now directed first into the baratte 
which was previously second, while the newly charged one is 
made the last of the series. The process is thus kept on con- 
tinuously. The apparatus used is shown in Fig. 2, 

Fig. 2. 

u. Charging pipe. b. Steam pipe. i. Steam drip. d. COj enters. / Discharge 
pipe. A. Agitator, made of iron rods. C. Tank in which the precipitate settles. 
B. Baratte body. D. Steam jacket. 

Each baratte holds about 1200 litres of solution, and the 
complete precipitation of all the alumina in it takes five to six 
hours. A mechanical agitator stirs the contents continually, 
and a current of steam is let into the double bottom so as to 


keep the temperature of the solution about 70°. The precipi- 
tated alumina and the solution of sodium carbonate which re- 
main are received in a vat placed beneath each baratte. The 
solution is decanted off clear, after standing, and then evapor- 
ated down to dryness, regenerating the sodium carbonate used 
in treating the bauxite to make the aluminate, less the inevit- 
able losses inseparable from all industrial operations. The de- 
posit of alumina contains considerable sodium carbonate me- 
chanically intermixed, and is put into a conical strainer to 
drain, or else into a centrifugal drying machine, which rapidly 
drives out of the hydrated alumina most of the solution of 
sodium carbonate which impregnates it ; a washing with pure 
water in the drier itself terminates the preparation of the 
alumina. At the works at Salindres, a part of this alumina is 
converted into sulphate of alumina, which is sold, the remainder 
being used for the aluminium manufacture. After washing in 
the drier, the alumina presents this composition : — 

Alumina 47.5 

Water 50.0 

Sodium carbonate 2.5 

Behnke* produces alumina by igniting bauxite or a similar 
mineral with sodium sulphate, carbon, and ferric oxide, using 
for each equivalent of alumina present at least one equivalent 
of alkali and one-half an equivalent of ferric oxide. The mix- 
ture is heated in a muffle or reverberatory furnace. The fritted 
product is ground, exposed to the air, and washed with water. 
Sodium aluminate goes into solution along with some sodium 
sulphate, while ferrous sulphide and undecomposed material 
remains as a residue. By passing carbonic acid gas or gases 
from combustion through the solution, the alumina is precip- 
itated. The residue spoken of is roasted, the sulphurous oxide 
given off utilized, and the residue used over in place of fresh 
ferric oxide. 

R. Lieber f proposes to treat bauxite, aluminous iron ore, 

* German Patent (D. R. P.), No. 7256. 
t German Patent (D. R. P.), No. 5670. 


etc., in a somewhat similar way. These materials are to be 
ground fine, mixed with sodium chloride and magnesium sul- 
phate (Kieserite), moistened with water, and pressed into 
bricks or balls. These are dried and put into a retort heated 
red-hot by generator gas. Hydrochloric acid gas is first given 
off, sodium sulphate and magnesium chloride being formed. 
In a further stage of the process sulphurous oxide is evolved, 
the alumina reacting on the sodium sulphate to form sodium 
aluminate. The latter is washed out of the residue, and its 
alumina precipitated by the ordinary methods. 

H. Muller* proposes to extract the alumina from silicates 
containing it by mixing them with limestone, dolomite, or 
magnesite, also with alkali caustic, carbonate, or sulphate (in 
the last case also with carbon), and heating the mixture to 
bright-redness. Alkaline aluminate is washed out of the re- 
sulting mass, while the residue, consisting of lime, magnesia, 
iron oxide, etc., is mixed wifh water-glass and moulded into 
artificial stone. 

Common salt is said not to react on bauxite if fused with it 
alone, but will decompose it if steam is used. Tilghmanf first 
used this reaction in 1847. It is said that it was also used at 
Nanterre and Salindres previously to 1865. A mixture of 
sodium chloride and bauxite was treated in a closed retort and 
steam passed through, or, better, in a reverberatory furnace an(J 
steam passed over it, at a high temperature. Much sodium 
chloride must have been lost by the latter arrangement. The 
fused mass was treated with water, when sodium aluminate dis- 
solved out. 

R. WagnerJ proposed to make a solution of sodium sulphide, 
by reducing sodium sulphate by carbon bisulphide, and to boil 
the bauxite in it. The sulphuretted hydrogen evolved was to 
be absorbed by ferric hydrate; while the sodium aluminate 

* German Patent (D. R. P.), No. 12,947. 
t Polytechnisches Journal, 106, p. 196. 
J Wagner's Jahresb., 1865, p. 332. 


was converted into soda and alumina by any of the ordinary 

Laur, a manufacturer of alumina in the south of France, has 
attempted to use sodium sulphate instead of the carbonate, in 
calcining bauxite. He finds it disadvantageous to exclude the 
carbonate altogether, but uses for a bauxite containing 20 per 
cent, of ferric oxide and 60 per cent, of alumina, about 66 of 
sodium sulphate and 16 of sodium carbonate per 100 of baux- 
ite. Carbon must also be mixed with the charge. Under 
these conditions the ferric oxide present is first reduced to iron, 
which finally becomes iron sulphide. The soda from both the 
carbonate and the sulphate combines with the alumina to form 
aluminate. An excess of soda is necessary to keep -the 
aluminate in solution, and to keep the sulphide of iron out of 
solution. The rest of the process is similar to that already de- 
scribdti at Salindres. 

According to Lowig's experiments, solution of sodium alumi- 
nate can be precipitated by calcium, barium, strontium, or 
magnesium hydrates, forming caustic soda and hydrated 
alumina, the latter being precipitated, together with lime, 
baryta, strontia, or magnesia. The precipitate is washed by 
decantation, and then divided into two portions, one of which 
is dissolved in hydrochloric acid, the other made into a mush 
with water, and gradually added to the solution of the first half 
until the filtrate shows only a very little alumina in solution. 
Chloride of calcium, barium, strontium, or magnesium has been 
formed, and the alumina all precipitated. 

Dr. K. J. Bayer has made an improvement in the process of 
extracting alumina from bauxite, which has received great 
commendation from those directly interested in the business, 
and who may be supposed to have proved its merits. Dr. 
Bayer thus describes it : * Bauxite is fused with sodium car- 
bonate or sulphate, and the solution obtained by washing, 
containing sodium aluminate, is not decomposed by carbonic 

* Stahl und Eisen, Feb. 1889, p. 112. 


acid, as formerly, but by the addition of aluminium hydrate 
with constant stirring. The decomposition of the solution goes 
on until the quantity of alumina remaining in solution is to the 
sodium protoxide as i to 6. This preciptation takes place in 
the cold, and the pulverulent aluminium hydrate separted out 
is easily soluble in acids. The alkaline solution remaining is 
concentrated by evaporation, taken up by ground bauxite, 
dried, calcined, and melted, and thus goes through the process 
again. The use of this caustic soda solution containing alumina 
is thus much more profitable than using soda, because by using 
the latter only 75 per cent, of the bauxite used is utilized, 
whereas by the former all th« alumina dissolved by the solution 
is obtained again. 

3. Preparation of Alumina from Cryolite. 

By the Dry Way. — The following method was invented by 
Julius Thomson ; the description is taken principally from Mier- 
zinski's " Fabrikation des Aluminiums :" The cryolite is pulver- 
ized, an easy operation, and to every lOO parts, 130 to 150 
parts of chalk are added, and a suitable quantity of fluorspar 
is also used, which remains in the residue on washing after ig- 
nition. More chalk is used than is theoretically necessary, in 
order to make the mass less fusible and keep it porous. But, 
to avoid using too much chalk merely for this purpose, a cer- 
tain quantity of coke may be put into the mixture. It is of the" 
first importance that the mixture be very intimate and finely 
pulverized. It is of greater importance that the mixture be 
subjected to just the proper well-regulated temperature while 
being calcined. The cryolite will melt very easily, but this is 
to be avoided. On this account, the calcination cannot take 
place in an ordinary smelting furnace, because, in spite of stir- 
ring, the mass will melt at one place or another, while at an- 
other part of the hearth it is not even decomposed, because 
the heat at the fire-bridge is so much higher than at the farther 
end of the hearth. Thomson constructed a furnace for this 
special purpose (see Figs. 3 and 4), in which the flame from 


1 45 

the fire first went under the bed of the furnace, then over the 
charge spread out on the bed, and finally into a flue over the 
roof of the hearth. The hearth has an area of nearly 9 square 
metres, being 4 metres long and 2.5 metres wide. It is charged 

Fig. 3. 


twelve times each day, each time with 500 kilos of mixture, 
thus roasting 6000 kilos daily, with a consumption of 800 kilos 
of coal. The waste heat of the gases escaping from the furnace 

Fig. 4. 


is utilized for drying the soda solution to its crystallizing point, 
and the gases finally pass under an iron plate on which the 
chalk is dried. In this furnace the mass is ignited thoroughly 
without a bit of it melting, so that the residue can be fully 
washed with water. 


The decomposition takes place according to the formula — 

2 (AlFs-sNaF) + 6CaC03=Al203.3Na20 + 6CaF, + 6C0„ 

the resultant product containing aluminate of soda, soluble in 
water, and insoluble calcium flouride. The reaction com- 
mences at a gentle heat, but is not completed until a red heat 
is reached. Here is the critical point of the whole process, 
since a very little raising of the temperature above a red heat 
causes it to melt. However, it must not be understood that the 
forming of lumps is altogether to be avoided. These lumps 
would be very hard and unworkable when cold, but they can be 
broken up easily while hot, so that they may be drawn out of 
the furnace a few minutes before the rest of the charge is re- 
moved, and broken up while still hot without any trouble. 
The whole charge, on being taken out, is cooled and sieved, the 
hard lumps which will not pass the sieve are ground in a mill 
and again feebly ignited, when they will become porous and 
may be easily ground up. However, the formation of these 
lumps can be avoided by industrious stirring of the charge in 
the furnace. A well-calcined mixture is porous, without dust 
and without lumps which are too hard to be crushed between 
the fingers. We would here remark that mechanical furnaces 
of similar construction to those used in the manufacture of soda, 
potash, sulphate of soda, etc., are more reliable and give the 
best results if used for this calcination. The mixture, or ashes, 
as the workmen call it, is drawn still hot, and washed while 
warm in conical wooden boxes with double bottoms, or the box 
may have but one bottom, with an iron plate about "]& milli- 
metres above it. A series of such boxes, or a large apparatus 
having several compartments, may be so arranged that the 
washing is done methodically, i. e., the fresh water comes first 
in contact with a residue which is already washed nearly clean, 
and the fresh charge is washed by the strong liquor. This is 
known as the " Lessiveur methodique," and an apparatus con- 
structed especially for this purpose is described in Dingier 186, 
376, by P. J. Havrez, but the subject is too general and the de- 


scription too long to be given here. A very suitable washing 
apparatus is also the Buff-Dunlop, used in the soda industry 
for washing crude soda, and described in Lunge's " Sulphuric 
Acid and Alkali," Book II., p. 465. Since the ashes are 
taken warm from the furnace the washing water need not be 
previously heated, but the final wash-water must be warmed, as 
the ashes have been cooled down by the previous washings. 
As soon as the strong liquor does not possess a certain 
strength, say 20° B., it is run over a fresh charge and so 
brought up. The solution contains sodium aluminate. 

The carbon dioxide necessary for precipitating the hydrated 
alumina may be made in different ways. The gases coming 
from the furnace in calcining the cryolite might be used if they 
were not contaminated with dust; and there is also the diffi- 
culty that exhausting the gases from the furnace would inter- 
fere with the calcination. It has also been recommended to 
use the gases from the fires under the evaporating pans, by 
exhausting the air from the flues and purifying it by washing 
with water. This can only be done where the pans are fired 
with wood or gas. However, the lime-kiln is almost exclu- 
sively used to furnish this gas. The kiln used is shaped like a 
small blast furnace. Leading in at the boshes are two flues 
from five fire-places built in the brickwork of the furnace, and 
the heat from these calcines the limestone. The gases are 
taken ofT by a cast-iron down-take at the top. At the bottom 
of the furnace, corresponding with the tap hole in a blast fur- 
nace, is an opening, kept closed, from which lime is withdrawn 
at intervals. A strong blast is blown just above the entrance 
of the side flues, and by keeping up a pressure in the furnace, 
leakings into it may be avoided. The gas is sucked away from 
the top by a pump, which forces it through a cleaning appar- 
atus constructed like a wash-bottle, and it is then stored in a 
gasometer. Instead of the pump, a steam aspirator may be 
used, which is always cheaper and takes up less room. 

The precipitation with carbonic acid gas is made by simply 
forcing it through a tube into the liquid. The apparatus used 


at Salindres is one of the most improved forms. (See p. 140.) 
The precipitate is granular, and settles easily. However, it is 
not pure hydrated alumina, but a compound of alumina, soda, 
carbonic acid, and water, containing usually about — 

Alumina ■ 45 per cent. 

Sodium carbonate 20 " 

Water 35 " 

The sodium carbonate must be separated by long-continued 
boiling with water, but by this treatment the alumina becomes 
gelatinous and diflficult of further treatment. The precipitate 
was formerly separated on linen filters, but centrifugal ma- 
chines are now preferred. The evaporated solution gives a 
high grade of carbonate of soda free from iron. The heavy 
residue which is left after the ashes have been lixiviated con- 
sists of calcium fluoride with small quantities of ferric oxide, 
lime, undecomposed cryolite, and aluminate of soda, and has 
not been utilized for any purpose. 

R. Biederman* states that if steam is passed over molten 
cryolite at a white heat, hydrofluoric acid gas and sodium 
fluoride are formed and driven over, while a white, pure crys- 
talline mass of alumina remains. 

Utilization of aluminous fluoride slags. — At Nanterre, De- 
ville used the following process for utilizing in one operation 
the slags from the aluminium manufacture and the residues 
from the sodium manufacture. 

" The slags from making aluminium contain 60 per cent, of 
sodium chloride and 40 per cent, of insoluble matter; the 
former can be removed by a single washing. The insoluble 
material is almost entirely aluminium fluoride, with a little 
alumina and undecomposed cryolite. When fluorspar is used 
as a flux, the sodium chloride in the slag is in part replaced by 
calcium chloride ; but, in general, all the fluorine in the slag is 
found combined with the aluminium, which shows the great 
affinity between these two elements. The residues left in the 

* Karl and Stohman, 4th Ed., p. 819. 


sodium retorts deteriorate quickly when exposed to the air, and 
contain ordinarily, according to my analysis-^ 

Carbon 20.0 

Carbonate of soda 14.5 

Caustic soda 8,3 

Sulphate of soda 2.4 

Carbonates of lime and iron 29.8 

Water 25.0 


" To utilize these two materials, 5 to 6 parts of the sodium 
residues are mixed carefully with one part of the washed slag, 
and the whole calcined at a red heat. The fusion becomes 
pasty ; it is cooled and washed, when aluminate of soda goes 
into solution, and on treatment with carbon dioxide gives 
sodium carbonate and alumina. According to my laboratory 
experiments — 

1000 grammes of sodium residues 
160 " " washed slags 

have given 

no grammes of calcined alumina 
225 " " dry sodium carbonate. 

" The residue left on washing the fusion weighs about one- 
half the weight of the soda residues used, and contains — 

Carbon 30.0 

Calcium fluoride 32.0 

Alumina 0.6 

Various other materials 37.4 

" The latter item is formed of ferric oxide, oxide of man- 
ganese, a little silica and some oxysulphide of calcium." 

Decomposition of Cryolite in the Wet Way. — Deville used the 
following method at Javel, which he thus describes; — 

" In the Greenland cryolite there are to be found numerous 
pieces containing siderite (ferrous carbonate). It is necessary 


to extract all these pieces before using the mineral as a flux in 
producing aluminium. The rejected fragments are then util- 
ized by pulverizing them finely, mixing with about three-fourths 
of their weight of pure, burnt lime and the whole carefully 
slaked. After the slaking, water is added in large quantity, 
and the material is heated in a large cast-iron vessel by means 
of a steam-coil. A reaction takes place at once, and is com- 
plete if the process is well conducted. Some insoluble alum- 
inate of lime may be formed, but it can be recovered from the 
residue by digesting itwith some solution of carbonate of soda. 
The residue remaining is calcium fluoride, which settles easily, 
and the clear liquor decanted off contains aluminate of soda, 
from which alumina can be precipitated as before. The cal- 
cined alumina obtained may contain iron when the cryolite 
used contains a large amount of ferrous carbonate. It has ap- 
peared to me that the latter mineral may be decomposed by 
the lime, and some protoxide of iron be thus dissolved by the 
soda in small quantity." 

"We make alumina by this method at Nanterre only because 
it utilizes the impure pieces of cryolite and works in conveni- 
ently with the previously-described processes for utilizing the 

An ingenious modification of the above process was devised 
by Sauerwein. The first reaction is the same, five parts of 
finely-powdered cryolite being boiled with four parts of burnt 
lime, as free as possible from iron, producing a solution of 
sodium aluminate and a residue of insoluble calcium fluoride. 
Tissier recommended using two parts' of cryolite to one of lime, 
but with these proportions only about one-third of the alumin- 
ium in the cryolite is converted into soluble aluminate. Hahn 
claims that complete decomposition takes place by using xoo 
parts of cryolite to 88 parts of burnt lime. The solution is set- 
tled, washed by decantation, and these washings put with the 
strong solution first poured off; the next washings are reserved 
for the fresh wash-water of another operation. The solution of 
sodium aluminate is then boiled with a quantity of cryolite 


equal to the amount first used, when sodium fluoride is formed 
and alumina precipitated. This operation is in no way diffi- 
cult, only requiring a little more attention than the first. 
The alumina thus made is very finely divided. The reactions 
involved are : 

2(AlF3.3NaF) + 6CaO = AlA-sNa^O + 6CaF,. 
2(AlFF3.3NaF) + AlA-sNa^O + 6H,0 = aCAlA-sH^O) + izNaF. 

During the last operation it is best to add an excess of cryo- 
lite, and keep the liquid in motion to prevent that mineral from 
caking at the bottom. Lead is the best material to make these 
precipitating tanks of, since iron would contaminate the alumina. 
The precipitate is washed as in the previous operation. The 
solution of sodium fluoride is boiled with the requisite quantity 
of burnt lime, which converts it into caustic soda, NaOH, which 
is separated from the precipitated calcium fluoride by decanta- 
tion and washing. 

In the establishment of Weber, at Copenhagen, where at one 
time all the cryolite produced in Greenland was received, the 
mineral was decomposed by acid. Hydrochloric acid attacks 
the mineral slowly, but sulphuric acid immediately dissolves 
the sodium fluoride, with disengagement of hydrofluoric acid ; 
gelatinous aluminium fluoride separates out, and is attacked 
more slowly. The cryolite requires nearly i ^ parts of sul- 
phuric acid to dissolve it, the reaction being : 

2(AlF3.3NaF) + 6H2SO4 = AlsCSO^s + 3Na,S04+ 12HF. 

The solution is evaporated and crystallized, when the sodium 
sulphate crystallizes out, and the mother liquid is treated for its 
alumina. This method is too costly, when compared with more 
recent processes, to be used at present. 

According to Schuch* very finely-powdered cryoHte is dis- 
solved by a large excess of hot dilute soda solution, but is 
thrown down unaltered when carbonic acid gas is passed through 

* Polytechnisches Journal, 165, p. 443. 


the solution. An excess of concentrated soda liquor converts 
the mineral into sodium aluminate and sodium fluoride, the for- 
mer being soluble but the latter almost insoluble in the soda 


The Preparation of Aluminium Chloride and 
Aluminium-Sodium Chloride. 

Anhydrous aluminium chloride cannot be made by evap- 
orating the solution of alumina in hydrochloric acid, for, as we 
have seen, decomposition of the salt sets in, hydrochloric acid 
is evolved and alumina remains. The same phenomena occur 
on evaporating a solution of the double chloride. These anhy- 
drous chlorides are prepared by a method discovered by 
Oerstedt, applicable to producing a number of similar metallic 
chlorides, which consists in passing a current of dry chlorine 
gas over an ignited mixture of alumina and carbon. 

Wohler* proceeded as follows in preparing the aluminium 
chloride which was used in his early experiments : " Alumina is 
mixed with charcoal powder and made plastic with oil. Cylin- 
ders of about 5 millimetres diameter are made of this paste, 
placed in a crucible with charcoal powder and heated until no 
more combustible gases distil. After cooling the crucible the 
cylinders are taken out, and a porcelain or glass tube open at 
both ends filled with them. This is then placed in a combus- 
tion furnace, connected at one end with a chlorine generator, 
and at the other with a tubular extension from the further end 
of which the gases escape, either into the air or into a flask 
filled with milk of lime. When the whole apparatus is ready, 
and filled with well-dried chlorine gas, the tube is heated to 
glowing, when aluminium chloride is formed and condenses in 
the extension of the tube." 

Deville paid great attention' to the production and purifica- 

*Pogg. Ann., n, p. 146. 



tion of aluminium chloride ; the following is his account of the 
processes used at Javel : 

Manufacture on a Small Scale. — " I took 5 kilos of alumina 
and mixed it with 2 kilos of carbon and a little oil ; the paste 
was made into balls and ignited at a bright-red heat. The 
compact, coke-like mass resulting was broken in pieces and 
put, with its powder, into a stoneware retort, C (Fig. 5), 
having a capacity of about 10 litres, and terminating in a neck. 

Fig. s. 

D. This retort was put in a furnace and heated to redness, 
while a current of dry chlorine gas passed in by the tube A. 
During the first few moments considerable quantities of water 
vapor escape from the neck. When aluminium chloride distils, 
as is shown by dense, white fumes, a porcelain or stoneware 
funnel, E, is adjusted to the neck D, and kept in place by fill- 
ing the joint with fine asbestos and then luting it over with a 
little potter's clay mixed with hair. Against this funnel fits a 
globular vessel, F, the joint being made tight in a similar way. 
This apparatus condenses and holds all the chloride distilled. 
However fast the chlorine may pass into the retort, it is so well 
absorbed during three-fourths of the operation that not a trace 
is mixed with the carbonic oxide escaping. However, the gas 
always fumes a little, because of a small quantity of silicon 
chloride being formed by the chlorine and carbon attacking the 
sides of the retort, or from chloride of sulphur or a little chlor- 


oxycarbonic acid. When the globe F is filled it is taken away 
to extract the coherent, crystalline aluminium chloride it con- 
tains, and is replaced immediately by another. During one 
operation three jars were thus filled, and altogether a little over 
lO kilos of chloride obtained. In the retort there remained 
almost a kilo of coke mixed with alumina in the proportion of 
two of carbon to one of alumina, making 330 grammes of the 
latter remaining unattacked out of S kilos. This coke contains 
also some double chloride of alumina and potassium and a little 
calcium chloride, which render it deliquescent. This residue 
was washed, mixed with a fresh quantity of alumina, and em- 
ployed in a new operation." 

Manufacture on a large scale. — " In applying this process on 
a large scale, the oil and carbon were replaced by tar, the 
alembic by a gas-retort, and the glass receiver by a small brick 
chamber lined with glazed tiles. The alumina was obtained by 
calcining ammonia alum in iron pots ; the residue obtained by 
one calcination at a bright-red heat was mixed with pitch, to 
which a little charcoal dust was added. The paste was well 
mixed, introduced into iron pots, covered carefully and heated 
until all vapors of tar ceased burning. The aluminous carbon 
is used while it is still hot, if possible, as it is quite hygroscopic. 
(This aluminous carbon conducts electricity wonderfully well ; 
it is the best electrode to use in making aluminium by the bat- 
tery, since its alumina regenerates the bath.) The residue is 
hard, porous, and cracked, and contains sulphur from the sul- 
phuric acid of the alum, a little iron, phosphoric acid in small 
quantity, a perceptible proportion of lime, and finally potash, 
which is always present in alums made from clay. The chlo- 
rine gas used was conducted by lead pipes and passed over 
calcium chloride before being used. The retort used was of 
about 300 litres capacity, and was placed vertically in a sort of 
chimney, C (Fig. 6), the flame circulating all around it. In 
the bottom was a square opening, x, about 20 centimetres 
square, which could be closed by a tile kept in place by a screw, 
V. A porcelain tube pierced the sides of the furnace and en- 



tered the retort at O; it was protected from the flame by a fire- 
clay cyHnder inclosing it. At the top, the retort was closed by 
a tile, Z, of refractory brick, in the centre of which was made 
a square opening, W, oi \o \.o \2 centimetres side. Finally, an 
opening, X, placed 30 centimetres below the plate Z, gave 
issue to the vapors distilled, conducting them into the chamber 
L. This condensation chamber was about i metre cube ; it had 

Fig. 6. 

one wall of bricks in common with the furnace, thus keeping 
it rather hot. The other walls should be thin and set with 
close joints and very little mortar. The cover, M, was mov- 
able ; it and the sides of the chamber were of glazed tiles. An 
opening 20-30 centimetres square in the lower part of the 
chamber communicated with flues lined with lead, for a little 
chloride was drawn into them. The uncondensed gas passed 
to a chimney. 

" To work such an apparatus it is necessary, first of all, to 
dry it with the greatest care in all its parts, especially the con- 
densation chamber. The retort is slowly heated and is left 


Open at Z until it is judged quite dry, and is then filled with 
red-hot, freshly-calcined mixture of carbon and alumina. The 
top cover is then replaced, and the fire urged until the retort is 
at a dark-red heat all over. Finally, chlorine is passed in, but 
the opening at W is kept open ; the gas is allowed to pass into 
the condensation chamber only when fumes of aluminium 
chloride appear very abundantly at W. When the operation 
proceeds right, almost all the aluminium chloride is found at- 
tached in a dense, solid mass to the cover M. I have taken 
out at one time a plate weighing almost 50 kilogrammes, which 
was less than 10 centimetres thick; it was made up of a large 
number of sulphur-yellow crystals penetrating each other and 
looking like stalactites and long soda crystals. When it is 
judged that the material in the retort is almost exhausted, the 
hole X is opened, the residue scraped out, and fresh mixture 
put in. During the operation there should be no white vapors 
coming from the condensation chamber, but the odor of gas 
will always be sharp because of the silicon chloride present, 
formed unavoidably by the chlorine attacking the retort. A 
gas retort, handled well, should last continuously two or three 
months, or even more. The furnace should be constructed so 
as to permit its easy replacement without much expense. 
When in use, the retort is closely watched through spy-holes in 
the wall, and any cracks which may appear promptly plastered 
up, if not large, with a mixture of fine asbestos and soda 

Purification of aluminium chloride. — " It often happens that 
the chloride obtained is not pure, either from the nature of the 
apparatus employed, or from neglect of the many precautions 
which should be taken. In this case, to purify it, it is heated 
in an earthen or cast-iron vessel with fine iron turnings. When 
the hydrochloric acid, hydrogen and permanent gases are 
driven from the apparatus, it is closed and heated hotter, which 
produces a light pressure under which influence the aluminium 
chloride melts and enters into direct contact with the iron. 
The ferric chloride, which is as volatile as aluminium chloride. 


is transformed into ferrous chloride, which is much less volatile, 
and the aluminium chloride can be obtained pure by being 
volatilized away or distilled in an atmosphere of hydrogen." 

When the processes just described were put in use at the 
chemical works at La Glaciere, great care had to be taken to 
avoid letting vapors and acid gases escape into the air, since 
the works were surrounded by dwellings. To avoid these in- 
conveniences, the vapor of aluminium chloride was made to 
enter a heated space in which was sodium chloride, in order to 
produce the less volatile double chloride; but the apparatus 
choked up so persistently that the attempt was given up. It 
then occurred to Deville to put sodium chloride into the mix- 
ture itself in the retort. The same apparatus was used as be- 
fore, except that the large gas-retort had to be replaced by a 
smaller earthen one, which could be heated much hotter, the 
grate being carried half way up the retort.* The condensation 
chamber had to be replaced by a small earthen vessel. The 
double chloride produced is fusible at about 200°, and is quite 
colorless when pure, but colored yellow by iron. It is, more- 
over, very little altered in dry air when in compact masses, and 
can be easily handled. When the double chloride is obtained 
quite pure, it gives up its aluminium completely when reduced 
by sodium. 

The following description by M. Margottetf will show the 
form of apparatus used in 1882 by the French company carry- 
ing on the Deville process at Salindres : — 

The double chloride may be obtained in the same manner as 
the simple chloride ; it is sufficient to put some common salt, 
NaCl, into a mixture of alumina and carbon, and on heating 
this mixture strongly there is formed, by the action of the 
chlorine, aluminium-sodium chloride, which distils at a red heat 
and condenses in a crystalline mass at about 200°. The hy- 

* It was when first using this process that Deville borrowed some zinc retorts from 
the Vielle Montagne works, and since they contained a little zinc in their composition 
the aluminium made for a while was quite zinciferous. 

t Fremy's Ency. Chimique. 



drated alumina obtained in the preceding operation is mixed 
witli salt and finely pulverized charcoal, in proper proportions, 
the whole is sifted, and a mixture produced as homogeneous 
as possible ; then it is agglomerated with water and made into 
balls the size of the fist. These balls are first dried in a drying 
stove, at about 150°, then calcined at redness in retorts, where 
the double chloride should commence to be produced just as 
the balls are completely dried. These retorts are vertical cyl- 
inders of refractory earth ; each one is furnished with a tube in 
its lower part for the introduction of chlorine, and with another 
towards its upper end for the exit of the vapor of double chlo- 
ride (see Fig. 7). A lid carefully luted during the operation 

Fig. 7. 

with a mixture of fine clay and horse-dung serves for the charg- 
ing and discharging of the retort. The double chloride is con- 
densed in earthen pots like flower-pots, made of ordinary clay, 
and closed by a well-luted cover, into which passes a pipe of 
clay to conduct the gas resulting from the operation into flues 
connected with the main chimney. Each retort is heated by a 
fire, the flame of which circulates all round it, and permits 
keeping it at a bright red heat. An operation is conducted as 
follows : The retort is filled with stove-dried balls, the lid is 


carefully luted, and the retort is heated gently till all the mois- 
ture is driven off. This complete desiccation is of great im- 
portance, and requires much time. Then chlorine, furnished 
by a battery of three generating vessels, is passed in. During 
the first hours, the gas is totally absorbed by the balls ; the 
double chloride distils regularly for about three hours, and runs 
into the earthen pots, where it solidifies. Toward the end, the 
distillation is more difficult and less regular, and the chlorine is 
then only incompletely absorbed. After each operation there 
remains a little residue in the retort, which accumulates and is 
removed every two days, when two operations are made per 
day. One operation lasts at least twelve hours, and a retort 
lasts sometimes a month. The double chloride is kept in the 
pots in which it was condensed until the time it is to be used 
in the next operation ; it is almost chemically pure, save traces 
of iron, and is easy to keep and handle. 

The following estimate was made by Wurtz, in 1872, show- 
ing the cost of a kilo of aluminium-sodium chloride as made 
by the above process: — 

Anhydrous alumina 0.59 kilos @ 86 fr. per 100 kilos = o it. 50.7 cent. 
Manganese dioxide 3.74 " " 14 " " " " = o " 52.3 " 
Hydrochloric acid 15.72 " " 3" " " " =0" 47.1 " 

Coal 25.78 " " 1.40 ' =0" 36.1 " 

Wages o" 23.8 " 

Expenses o " 38.0 " 

Total 2 fr. 48.0 cent. 

This is equal to about 22^ cents per pound. An average 
of 10 kilos of this was used to produce one kilo of aluminium, 
which shows a yield of only 70 per cent, of the contained alu- 
minium, and an increased cost of 6y cents on every pound of 
aluminium from the imperfection of reduction. In this respect 
there certainly seems large room for improvement. 

The largest plant ever erected for the manufacture of alu- 
minium-sodium chloride was that of the Aluminium Co. Ltd., 
at Oldbury, near Birmingham, England. The plant was com- 
menced in the latter part of 1887, and was in working order in 


July, 1888. The process is in principle indentical with that used 
at Salindres, but the whole was on such a large scale that the 
description is still interesting. 

Twelve large regenerative gas furnaces are used, in each of 
which are placed five horizontal fire-clay retorts about 10 feet 
in length, into which the mixture is placed. These furnaces are 
in two rows, of six each, along each side of a building about 
250 feet long, leaving a clear passage down the centre 50 feet 
wide. Above this central passage is a platform swung from 
the roof, which carries the large lead mains to supply chlorine 
to the retorts ; opposite each retort is a branch pipe controlled 
by a valve. The valves are designed so that the chlorine must 
pass through a certain depth of (non-aqueous) liquid, thus 
regulating the flow and preventing any back pressure in the re- 
tort from forcing vapor into the main. The opposite or back 
ends of the retorts are fitted with pipes which convey the vapor 
of the double chloride into cast-iron condensers and thence into 
brick chests or boxes, the outsides or ends of which are closed 
by wooden doors fitting tightly. Convenient openings are ar- 
ranged for clearing out the passages, which may become 
choked because of the quickness with which the double chlor- 
ide condenses. On looking down the centre of the building it 
presents the appearance of a double bank of gas retorts for 
making ordinary illuminating gas, except that the retorts are 
only one-high. (Fig. 8). 

The chlorine plant is on a correspondingly large scale, the 
usual manganese-dioxide method being employed and the spent 
liquor regenerated by Weldon's process. The chlorine gas is 
stored in large gasometers from which it is supplied to the re- 
torts at a certain pressure. The mixture for treatment is made 
by mixing hydrated alumina with common salt and carbon in 
the form of charcoal powder or lamp-black. This being well 
mixed is moistened with water, thrown into a pug-mill from 
which it is forced out as solid cylinders, and cut into about 3 
inch lengths by a workman. The lumps are then piled on top 
of the large chloride furnaces to dry. In a few hours they are 



hard enough to allow handling, and are put into large wagons 
and wheeled to the front of the retorts. 

When the retorts are at the proper temperature for charging, 
the balls are thrown in until the retort is quite full, the fronts 
are then put up and luted tightly with clay, and the charge left 
alone for about four hours, during which the water of the hy- 
drated alumina is completely expelled, the rear end of the re- 
tort being disconnected from the condensing chamber, which 

Fig. 8. 

must be kept perfectly dry, and connected directly with the 
chimney. At the end of this time the chlorine is turned on 
and the retort connected with the receiver. At first the chlo- 
rine passed in is all absorbed by the charge and only carbonic 
oxide escapes into the boxes, where it is ignited and burns, 
thus warming them up. After a certain time dense fumes are 
evolved, and then the condensers are shut tightly and the un- 
condensed gases pass into the chimney. The chlorine is passed 
in for 72 hours in varying quantity, the boxes at the rear being 
opened from time to time by the workmen to note the pro- 
gress of the distillation. The greater part of the double chlo- 


ride liquefies and trfckles down to the floor of the chambers, 
but a portion sublimes and condenses on the walls as a yellow 
crystalline powder. These chambers are emptied from time to 
time and the contents packed away in air-tight wooden chests, 
that it may keep without absorbing moisture from the air. At 
the end of the distillation the chlorine valves are closed and 
the condenser boxes cleaned out ; the retorts are also opened 
at their front end and the residue raked out. This residue 
consists of a small quantity of alumina, charcoal and salt, and 
is remixed in certain proportions with fresh material and used 
over again. The retorts are then immediately re-charged and 
the operations repeated. Each set of five retorts produces 
about 1600 to 1800 lbs. in one operation, or say 3500 lbs. per 
week. The twelve furnaces are therefore capable of producing 
easily 1,500,000 lbs. of double chloride per annum. Since 10 
lbs. of this salt are required to produce i lb. of aluminium, the 
capacity of the works is thus seen to be 150,000 lbs. or over of 
metal per year. 

This last remark as to the proportion of chloride required to 
form the metal will show the absolute necessity there is to keep 
iron from contaminating the salt. This gets in, in varying pro- 
portions, from the iron in the materials used and in the fire-clay 
composing the retort, and exists as ferrous and ferric chlorides. 
Exercising the utmost care as to the purity of the alumina and 
charcoal used, and after having the retorts made of a special 
fire-clay containing a very small percentage of iron, it was found 
impossible to produce a chloride on a large scale containing less 
than 0.3 per cent, of iron. This crude chloride is highly de- 
liquescent and varies in color from light yellow to dark red — 
the color depending not so much on the absolute amount of 
iron present as on the proportion of iron present as ferric salt, 
which has a high color. Since practically all the iron present 
in the salt passes into the aluminium, it is seen that the latter 
would contain 3 per cent, or more of iron. For some time the 
only way to obviate this difficulty was to resort to purifying the 
aluminium, by which the content of iron was finally reduced to 


2 per cent. Mr. Castner has since perfected a process for 
purifying the double chloride by which only o.oi per cent, of 
iron is left in it. The principle employed in doing this is de- 
scribed in the patent claims* to be the reduction of the iron salts 
to metallic iron by melting the chloride (single or double) with 
a quantity of metallic aluminium or sodium sufiScient for this 
purpose. The purified chloride is quite white and far less de- 
liquescent than the crude salt, which seems to indicate that the 
iron chlorides have a large share in rendering the crude salt so 
deHquescent. The purified chloride is preserved by melting 
and running into tight iron drums. A process of purification 
used later by Mr. Castnerf consisted in separating out the iron 
electrolytically. The melted salt was passed slowly along a 
trough on the sides of which were electrodes kept at a tension 
sufficient to decompose the iron chlorides, the tension also 
being gradually decreased from one end to the other propor- 
tionally to the decreasing quantity of iron in the material as it 
passed along. 

The success of the manufacture of the double chloride is said 
to depend on the proportions of the mixture, the temperature 
of the furnace, the quantity of chlorine introduced, and the de- 
tails of construction of the retorts ; but very little information 
on these points has been made public. The following figures 
may give some idea of the quantities of materials used : The 
production of 1000 lbs. of double chloride is said to require — 

Common salt 357 lbs. 

Hydrated alumina 491 " 

Chlorine gas 674 " 

Coal 1800 " 

The salt and hydrated alumina are therefore mixed in about 
the same proportion as those indicated by the formula which 
represents the reaction 

*U. S. Patent, No. 409,668, Aug. 27, 1889. 
tU. S. Patent, No. 422,500, March 4, 1890. 



for if we assume the hydrated alumina used to contain 90 per 
cent, of that compound, the 491 lbs. of it used would corres- 
pond to very nearly the amount of salt said to be used. As 
to the cost of this double chloride, so many uncertain elements 
enter into it that it cannot be satisfactorily estimated from the 
data at hand. We are informed, however,* that the double 
chloride used represented 43 per cent, of the cost of aluminium 
to this company. If we place the total cost at 8 shillings per 
lb. this would indicate a trifle over 4 pence per lb. as the cost 
of the double chloride. It was probably not over 3 pence. 

H. A. Gadsden,! of London, patented a method of obtaining 
aluminium, in which the aluminium chloride used is obtained by 
a method similar in all respects to the process as described by 
Deville, except that the corundum or bauxite used is mixed 
with about 10 per cent, of sodium or potassium fluoride and a 
small quantity of fluorspar. After this has been mixed and 
calcined it is pulverized, 10 per cent, of charcoal dust added, 
made into balls and heated in a muffle until pasty. Taken 
out of the muffle they are then put into a retort, heated highly, 
and chlorine gas passed over them, when aluminium chloride 

Count R. de Montgelas| patents a process for producing 
aluminium chloride and the double chloride with sodium, in • 
which the only difference from the preceding methods is that 
molasses is used instead of pitch for moulding the mixture into 
balls, the mixture otherwise containing alumina, charcoal, and 
sodium chloride ; and it is claimed that by regulating the heat 
at which chlorine is passed over this mixture, previously cal- 
cined, aluminium chloride can be volatilized while aluminium- 
sodium chloride remains in the retort. The use of horizontal 
retorts is recommended. 

*Zeitschrift des Verein Deutscher Ingenieure, 1889, p. 301. 

t German Patent (D. R. P.) No. 27,572 (1884). 

I English Patents Nos. 10,011, 10,012, 10,013, Aug. 4, 1886. 


Prof. Chas. F. Mabery,* of the Case School of Applied Sci- 
ence, Cleveland, patented and assigned to the Cowles Bros, the 
process of making aluminium chloride, consisting in passing 
dry chlorine or hydrochloric acid gas over an alloy of alumin- 
ium and some other metal kept in a closed vessel at a temper- 
ature sufficient to volatilize the aluminium chloride formed, 
which is caught in a condenser. Or, hydrochloric acid gas is 
passed through an electrically heated furnace, in which alumina 
is being decomposed by carbon, a condenser being attached to 
the opposite end of the furnace. 

Mr. Paul Curiej states that aluminium chloride can be made 
by passing vapors of carbon disulphide and hydrochloric acid 
either simultaneously or successively over ignited alumina or 
clay. The first forms aluminium sulphide, which the latter 
converts into the volatile chloride, which distils. 

H. W. WarrenI recommends the following process as of gen- 
eral application in producing anhydrous metallic chlorides. 
Petroleum is saturated with either chlorine or hydrochloric acid 
gas, both gases being soluble in it to a large extent, particu- 
larly the latter gas. This operation is performed at a low tem- 
perature, as more of the gases is then dissolved. The oxide of 
the metal, alumina, for instance, is put into large earthenware 
retorts, and raised to red heat. The saturated oil is then boiled 
and its vapor passed into the retort. On contact with the 
oxide a strong reaction commences, fumes of aluminium chlor- 
ide are at once evolved, and distil into a condenser, the opera- 
tion being continued until no more white fumes appear. Then 
fresh alumina is supplied, and the reaction continues. The 
aluminium chloride may be purified from any oil by gentle ap- 
plication of heat. Mr. Warren also used naphthaline chloride 
with advantage, as also chloride of carbon, but their high price 
rendered them unable to compare with petroleum in economy. 

* U. S. Patent, Oct. 26, 1886. 
t Chemical News, 1873, p. 307. 
J Chemical News, April 29, 1887. 


Aluminium bromide can be similarly prepared by substituting 
bromine for chlorine. 

Camille A. Faure, of New York, the well-known inventor of 
the Faure storage battery, has patented* a process for produc- 
ing aluminium chloride, which is very similar to the above 
method. The manipulation is described as follows : An oxy- 
genated ore of aluminium is brought to about a red heat by 
bringing it, in a furnace, into direct contact with the flame. 
When at proper heat the flame is cut off and a gas containing 
carbon and chlorine introduced. A mixture of petroleum vapor 
or a similar hydrocarbon with hydrochloric acid gas is pre- 
ferred. Vaporized chloride of aluminium immediately passes 
off into a condenser. 

In a paper written by Mr. Faure, and read before the French 
Academy of Sciences by M. Berthelot,t it was stated that the 
aim of this process was to suppress the prominent disadvant- 
ages of the older methods : viz., cost and wear and tear of re- 
torts, great consumption of fuel, slowness of the operation, 
large amount of labor, and cost of the chlorine. For this pur- 
pose the chlorine is replaced by hydrochloric acid gas and the 
carbon by a hydrocarbon. Since all pure hydrocarbons are 
decomposed at a red heat with deposition of carbon, the pro- 
cess would appear impracticable ; but a proper mixture of hy- 
drochloric acid gas and naphthaline vapor is said not to de- 
compose by the highest temperature alone, a new compound 
being formed, a sort of napthaline chloride, which is exceed- 
ingly corrosive and powerful enough to attack any oxide and 
convert it into chloride. To carry out the process a gas fur- 
nace with large bed is used. On this is spread a layer of small 
pieces of bauxite about two feet deep. The flame comes in 
over the ore, passes downward through it and through numer- 
ous holes arranged in the hearth, and thence to a chimney. In 
this way the heat of the gases is well utilized, while the layer 

* U. S. Patent, No. 385,345, July 3, 1888. 
t July 30, 1888. 


of bauxite is heated to whiteness on top and to low-red at the 
bottom. The flames are then turned off and the mixture of 
naphthaHne and hydrochloric acid vapors passed upward 
through the bed, and by their reaction producing aluminium 
chloride, which is diverted by suitable flues into a condenser. 
It is claimed that by careful fractional condensation the chlor- 
ides of silicon, iron, calcium, etc., formed from impurities in 
the bauxite, can be easily separated, that of silicon being 
more volatile and those of iron and calcium less volatile than 
aluminium chloride. As naphthaline is a bye-product from 
gas-works, it is claimed that it can be bought for i ^ cents per 
lb., and that only j^ oi a lb. is used per lb. of aluminium chlo- 
ride produced. It is also claimed that one furnace, with two 
men to work it, will produce 4000 lbs. of chloride a day. The 
estimated cost of the chloride is about i }4 cents per pound, of 
which 17 per cent, is for beauxite, 47 per cent, for hydrochloric 
acid, 27 per cent, for naphthaline, and 9 per cent, for labor. 
Mr. Faure experimented in the vicinity of New York during 
1889, and was sanguine of having the process at work in 1890, 
but no commercial process has resulted from these efforts. 
(See further. Chap. XI.) 

In all the processes for producing aluminium chloride so far 
considered, the use of common clay was not recommended, 
since silicon chloride is formed as well as aluminium chloride. 
The only method proposed for using clay for this purpose is 
that of M. Dullo, nearly twenty years ago, and which cannot 
have been very successful, since it has not been heard of in 
operation. We will repeat his remarks, however, for there is 
still a large field open in the utilization of clay for the manu- 
facture of aluminium, and since the metal is becoming so cheap 
the manufacturers are not above looking for and utilizing the 
cheapest raw material available. 

* " Aluminium chloride may be obtained easily by direct 
treatment of clay. For this purpose a good clay, free from iron 
and sand, is mixed with enough water to make a thick pulp, 

* Bull, de la Soc. Chem., i860, vol. v., p. 472. 


to which are added sodium chloride and pulverized carbon. 
For every lOO parts of dry clay there are taken I20 parts of 
salt and 30 of carbon. The mixture is dried and broken up 
into small fragments, which are then introduced into a red-hot 
retort traversed by a current of chlorine. Carbonic oxide is 
disengaged, while at the same time aluminium chloride and a 
little silicon chloride are formed. It is not necessary that the 
chlorine should be absolutely dry, it may be employed just as 
it comes from the generator. The gas is absorbed very rapidly, 
because between the aluminium and silicon there are reciprocal 
actions under the influence of which the chemical actions are 
more prompt and energetic. The aluminium having for chlo- 
rine a greater afifinity than silicon has, aluminium chloride is 
first formed, and it is only when all the aluminium is thus trans- 
formed that any silicon chloride is formed. When the latter 
begins to form the operation is stopped, the incandescent mix- 
ture is taken out of the retort and treated with water. The 
solution is evaporated to dryness to separate out a small quan- 
tity of silica which is in it, the residue is taken up with water, 
and aluminium-sodium double chloride remains when the 
filtered solution is evaporated to dryness." 

We must say of M. Dullo's suggestions that it is the general 
experience that the more volatile silicon chloride is formed 
first; it is also very improbable that a solution of aluminium- 
sodium chloride can be evaporated without decomposition. 


The Preparation of Aluminium Fluoride and 
Aluminium-Sodium Fluoride (Cryolite). 

Natural cryolite is too impure for use in many operations 
which aim to produce very pure aluminium. Schuh has pro- 
posed boiling the mineral in solution of soda. Under certain 
conditions sodium aluminate is formed (see p. JJI), but if the 
solution of soda is dilute the liquor remains clear after taking 
up the cryolite, and on passing a current of carbonic acid gas 


through it aluminium-sodium fluoride is precipitated. In this 
way the pure double fluoride can be separated from impure 

Berzelius recommended preparing artificial cryolite by de- 
composing aluminium hydrate by a solution of sodium fluoride 
and hydrofluoric acid, the hydrate being added to the liquid 
until its acidity was just neutralized : — 

A1,03.3H,0 + 6NaF + 6HF = 2(AlF3.3NaF) + 6H,0. 

Deville states that on adding sodium chloride to a solution 
obtained by dissolving alumina in an excess of hydrofluoric 
acid, a precipitate of cryolite is obtained. Since cryolite is 
hardly attacked at all by hydrochloric acid, it is probable that 
the reaction occurring is 

AIF3 + 3HF + sNaCl = AlFa.sNaF + 3HCI. 

The process which Deville recommended as best, however, is 
the treatment of a mixture of calcined alumina and carbonate 
of soda, mixed in the proportions in which their bases exist in 
cryolite, by an excess of pure hydrofluoric acid : — 

AliiOa -I- sNa/'Os + i2HF=2(AlF.3NaF) -|- 3CO2 + 6H2O. 

100 parts of pure alumina requiring 310 parts of sodiurri car- 
bonate and 245 of anhydrous hydrofluoric acid, there being 
410 parts of cryolite formed. On drying the mass and melting 
it there results a limpid, homogeneous bath having all the 
characteristics of cryolite, being reduced by sodium or by an 
electric current, which would not result from a mere mixture ot 
alumina and sodium fluoride melted together. 

Deville also states that when anhydrous aluminium chloride 
is heated with sodium fluoride in excess, a molten bath results 
of great fluidity, and on cooling and dissolving away the excess 
of sodium fluoride by repeated washings the residue is similar 
to cryolite, while the solutions contain no trace of any soluble 
aluminium salt: — 

AICI3 + 6NaF=AlF3.3NaF 4- sNaCl. 


It is evident, however, that the above reaction would be the 
reverse of a profitable one, and is therefore not of economical 

Pieper* patents a very similar reaction, but operates in the 
wet way. A solution of aluminium chloride is decomposed by 
adding to it a suitable quantity of sodium fluoride in solution. 
Sodium chloride is formed and cryolite precipitated, as in the 
last reaction given. By adding different proportions of sodium 
fluoride solution, precipitates of double salts are obtained, con- 
taining varying proportions of the two fluorides. The use of 
aluminium chloride in solution would dispense with the objec- 
tion made to Deville's analogous method, and this process 
would very probably produce cryolite quite cheaply. 

Brunerf produced aluminium fluoride by passing hydro- 
fluoric acid gas in the required quantity over alumina heated 
red hot in a platinum crucible: — 

Al20, + 6HF=2A1F3 4 3H2O. 

Devillef states that it can be made by melting together the 
equivalent quantities of cryolite and aluminium sulphate : 

2(AlF3.3NaF) -I- Al,(S04)3.:eH,0 = 4AIF3 + sNa^SO^ + ;i;H.,0. 

On washing the fusion, sodium sulphate goes into the solu- 
tion. It is also stated that hydrochloric acid gas acting on a 
mixture of fluorspar and alumina at a high temperature will 
produce aluminium fluoride : 

AlA + sCaF^ + 6HC1 = 2AIF3 4- aCaCl, -1- 3H.,0. 

The calcium chloride would be partly volatilized and the re- 
mainder washed out of the fusion. 

Hautefeuille§ obtained crystallized aluminium fluoride by 

♦German Patent (D. R. P.), No. 35,212. 

t Pogg- Annalen, 98, p. 488. 

t Ann. de. Chim. et de Phys. [3], 61, p. 333; [3], 49, p. 79. 

§Idem, [4], 4.P- iS3- 


passing hydrofluoric acid gas and steam together over red-hot 

Ludwig Grabau, of Hanover, bases his process of producing 
aluminium on the reduction of aluminium fluoride (see Chap. 
X.), which is prepared on a commercial scale by the following 
ingenious methods : 

* The process is based on the conversion of aluminium sul- 
phate into fluoride by reaction with cryolite, the fluoride being 
afterward reduced by sodium in such a manner that a double 
fluoride of sodiuni and aluminium results, which is used over 
again, thus forming a continuous process. The purest obtain- 
able cryolite is used to start the process, after which no more 
is needed, the material supplying the aluminium being its sul- 
phate, which can be obtained cheaply in large quantities, and 
almost perfectly pure. The process is outlined by the reac- 
tions — 

2 (AlFa-sNaF) + AUCSO*), = 4AIF3 + sNa.SO^ 
2AIF3 + sNa = AH AlFs.sNaF. 

It is thus seen that theoretically the cryolite would be exactly 
reproduced, but the losses and incomplete reactions unavoida- 
ble in practice would cause less to be obtained, and necessitate 
the continual addition of fresh cryolite ; since, however, it is 
not desired to base the process on the continual use of cryo- 
lite, because of the impurities in that mineral, an indirect pro- 
cess is used, consisting of two reactions, in place of the first 
given above, in which theoretically a larger quantity of cryolite 
is finally obtained than is used to begin with. This is operated 
by introducing fluorspar into the process, the base of which 
goes out as calcium sulphate or gypsum, and so supplies the 
fluorine needed. 

In practice, a solution of aluminium sulphate is heated with 
powdered fluorspar (obtained as pure as possible and further 
cleaned by treatment with dilute hydrochloric acid). The 

♦German Patent (D. R. P.), No. 48,535. March 8, 1889. 


aluminium sulphate will not be entirely converted into fluoride, 
as has been previously observed by Friedel,* but about two- 
thirds of the sulphuric acid is replaced by fluorine, forming a 
fluorsulphate of aluminium. This latter compound remains in 
solution, while gypsum and undecomposed fluorspar remain as 
a residue and are filtered out. 

AlaCSO,)^-!- 2CaF, = Al^F.CSOi) +2 CaSOi. 

This solution is concentrated and mixed with cryolite in such 
proportion that the alkali in the latter is just equivalent to the 
sulphuric acid in the fluor-sulphate. The mass is dried and 
gnited, and the product washed and dried. 

3AUF,(S0,) + 2(AlF3.3NaF) =8A1F3+ 3Na,S04. 

On reduction with sodium, the 8 molecules of aluminium fluor- 
ide, treated with sodium as by the reaction given, produce 4 
molecules of the double fluoride. It is thus seen that after 
allowing for reasonable losses in the process there is much . 
more cryolite produced than is used, and the excess can be 
very profitably sold as pure cryolite, being absolutely free from 
iron or silica. 

Grabau proposed to obtain his aluminium sulphate solution 
directly from kaolin, by treatment with sulphuric acid, and thus 
utilize this most abundant of the aluminous minerals. It is 
claimed that aluminium fluoride of the greatest purity can be 
thus manufactured at a less cost than pure alumina can be 

More recently the method of production has been varied as 
follows:! To produce aluminium fluoride free from silicon and 
iron, calcined clay or kaolin is stirred in excess into dilute hy- 
drofluoric acid, a 12 per cent, solution of which is recommended. 
The violent reaction ensuing is moderated by cooling, so that 
the temperature does not exceed 95°. After a few minutes the 

* Bull, de la Soc. Chimique [2], xxi, 241. 

t German Patent No. 69,791, August 20, 1892. 


liquor is neutralized, as may be tested by a drop giving a pure 
yellow with tropaolin. This neutrality is essential to ensure 
the solution being free from silica. Further, this neutrality 
cannot be obtained using uncalcined clay, nor with clay which 
has been too highly heated. The clay must be calcined at a 
certain determined temperature, which can only be learned by 
experience. The hot, neutral solution is cooled to a moderate 
temperature and filtered quickly, washing the residue with hot 
water. On an average the equivalent of 95 per cent, of the hy- 
drofluoric? acid used can be obtained in the form of dissolved 
aluminium fluoride. The solution will be free from dissolved 
silica, but may contain iron, lead, arsenic or other accidental 
impurities. In order to separate out pure aluminium fluoride 
from this solution, Grabau has patented the following method of 
procedure :* The solution is first treated with sulphuretted hy- 
drogen gas, which precipitates any lead, copper, arsenic, etc., 
which may be present, and reduces any ferric salt present to the 
ferrous state. Any other reducing agent could also be used to 
effect this. The solution is filtered and slightly acidified, so 
that a test drop turns red with tropaolin. A neutral solution 
holding hydrogen sulphide would in the subsequent cooHng de- 
posit traces of iron sulphide. The acidified solution is then 
str-ongly cooled in a receptacle, which may be of sheet alumin- 
ium. There at once separates out crystalline, hydrated alu- 
minium fluoride, AIF3.9H2O. It is best to start the crystalliza- 
tion by dropping in a trace of the crystallized salt. As soon as 
the crystallization commences the temperature rises ; when it 
again sinks to 0° by continuing the cooling, the crystallization 
is ended. The thick crystal liquor is separated on filters or in 
a centrifugal machine, into mother liquor and crystals; the 
latter are washed with ice-cold water. These crystals are free 
from iron, but if the solution had not been previously reduced, 
noticeable quantities of ferric fluoride would have been in them. 
These crystals are easily dried and the water driven off by 
gentle heating. 

♦German Patent No. 70,155, August 20, 1892. 



The Preparation of Aluminium Sulphide. 

Until the researches of M. Fremy, no other method of pro- 
ducing aluminium sulphide was known save by acting on the 
metal with sulphur at a very high heat. Fremy was the first 
to open up a diflferent method, and it may be that his discov- 
eries will yet be the basis of successful industrial processes. In 
order to understand just how much he discovered, w? here give 
all that his original paper contains concerning this sulphide.* 

"We know that sulphur has no action on silica or boric ox- 
ide, magnesia, or alumina. I thought that it might be possible 
to replace the oxygen by sulphur if I introduced or intervened 
a second affinity, as that of carbon for oxygen. These decom- 
positions produced by two affinities are very frequent in chem- 
istry ; it is thus that carbon and chlorine, by acting simultane- 
ously on silica or alumina, produce silicon or aluminium 
chloride, while either alone could not decompose it ; a similar 
case is the decomposition of chromic oxide by carbon bisul- 
phide, producing chromium sesquisulphide. Reflecting on 
these relations, I thought that carbon bisulphide ought to act 
at a high heat on silica, magnesia, and alumina, producing 
easily their sulphides. Experiment has confirmed this view. 
I have been able to obtain in this way almost all the sulphides 
which until then had been produced only by the action of sul- 
phur on the metals. 

" To facilitate the reaction and to protect the sulphide from 
the decomposing action of the alkalies contained in the por- 
celain tube which was used, I found it sometimes useful to mix 
the oxides with carbon and to form the mixture into bullets 
resembling those employed in the preparation of aluminium 
chloride. I ordinarily placed the bullets in little carbon boats, 
and heated the tube to whiteness in the current of vaporized 

* Ann. de Chim. et de Phys. [3], 38, p. 312. 


carbon bisulphide. The presence of divided carbon does not 
appear useful in the preparation of this sulphide. 

" The aluminium sulphide formed is not volatile ; it remains 
in the carbon boats and presents the appearance of a melted 
vitreous mass. On contact with water it is immediately de- 
composed : 

A1,S, + 3H,0 = Al,03 + 3H,S. 

" The alumina is precipitated, no part of it going into solu- 
tion. This precipitated alumina is immediately soluble in weak 
acids. The clear solution, evaporated to dryness, gives no 
trace of alumina. It is on this phenomenon that I base the 
method of analysis. 

" Aluminium sulphide being non-volatile, it is always mixed 
with some undecomposed alumina. It is, in fact, impossible to 
entirely transform all the alumina into sulphide. I have heated 
less than a gramme of alumina to redness five or six hours in 
carbon bisulphide vapor, and the product was always a mixture 
of alumina and aluminium sulphide. The reason is that the 
sulphide, being non-volatile and fusible, coats over the alumina 
and prevents its further decomposition. The alumina thus 
mixed with the sulphide, and which has been exposed to a red 
heat for a long time, is very hard, scratches glass, and is in 
grains which are entirely insoluble in acids. By reason of this 
property I have been able to analyze the product exactly, for 
on treating the product with water and determining on the one 
hand the sulphuretted hydrogen evolved, and on the other the 
quantity of soluble alumina resulting, I have determined the 
two elements of the compound. One gramme of my product 
contained 0.365 grm. of aluminium sulphide, or 36.5 per cent., 
the remainder being undecomposed alumina." 

The composition of this sulphide was— 

Aluminium °-i37 g™- = 37-5 pef cent. 

Sulphur 0.228 "=62.5 

0.365 " lOO.O " 


The formula AI2S3 requires — 

Aluminium 36.3 per cent. 

Sulphur 63.7 " 

The above is the substance of Fremy's investigations and 
results. Reichel* next published an account of further experi- 
ments in this line. He found that by melting alumina and sul- 
phur together no reaction ensued. In the case of magnesia, the 
sulphide was formed if carbon was mixed with the magnesia 
and sulphur, but this change did not alter the alumina. Hydro- 
gen gas passed over a mixture of alumina and sulphur likewise 
gave negative results. Sulphuretted hydrogen passed over ig- 
nited alumina did not succeed. By filling a tube with pure alu- 
mina, passing in hydrogen gas and the vapor of carbon bisul- 
phide, the heating being continued until carbon bisulphide 
condensed in the outlet tube, and then hydrogen being passed 
through until the tube was cold, a product was obtained con- 
taining aluminium sulphide and undecomposed alumina. 

In 1886, the writer made a series of experiments on the pro- 
duction and reduction of aluminium sulphide. Alumina, either 
alone or mixed with carbon or with carbon and sulphur, was 
put in porcelain or carbon boats into a hard glass or porcelain 
tube. This was then heated and vapor of carbon bisulphide 
passed through it. The product was analyzed according to 
Fremy's directions. The proportion of aluminium sulphide 
obtained in the product varied from 13 to 40 per cent. The 
best result was obtained at the highest heat — almost whiteness. 
The presence of sulphur or carbon, or both together, mixed 
with the alumina, did not promote to any degree the formation 
of a richer product. The conditions for obtaining the best re- 
sults seem to be high heat and fine division of the alumina to 
facilitate its contact with the carbon bisulphide vapor. The 
product was light-yellow when not mixed with carbon, easily 
pulverized, and evolved sulphuretted hydrogen gas energetically 

* Jahresb. der Chemie, 1867, p. 155. 


when dropped into water. Since carbon bisulphide can now be 
manfactured at a very low price, say 2 to 3 cents per lb., it is 
not impossible that it may be found profitable to produce 
aluminium from its sulphide. In such a case, large retorts 
would be used, a stirring apparatus would facilitate the forma- 
tion of a richer product, and the unused carbon bisulphide 
could be condensed and saved. 

M. Comenge,* of Paris, proposed to prepare aluminium sul- 
phide by using a clay retort similar to those used in gas-works, 
filling it one- half its length with charcoal or coke and the other 
half with alumina. The retort being heated to rednesss, sul- 
phur is introduced at the coke end, when in contact with the 
carbon it forms carbon bisulphide, which acts upon the alumina 
at the other end, producing the sulphide. 

Messrs. Reillon, Montague, and Bourgerel f obtained a pat- 
ent in England for producing aluminium, in which aluminium 
sulphide is obtained by mixing powdered alumina with 40 
per cent, of its weight of charcoal or lampblack and formed 
into a paste with a sufficient quantity of oil and tar. This is 
then calcined in a closed vessel and an aluminous coke ob- 
tained. This is broken into pieces, put into a retort, and 
treated with carbon bisulphide vapor. The inventors state 
that the reaction takes place according to the formula 

2 A1,0, + 3C + 3CS, = 4A1,S, -^ 6C0. 

PetitjeanI states that if alumina is mixed with tar or turpentine 
and ignited in a carbon-lined crucible, and the coke obtained 
mixed intimately with sulphur and carbonate of soda and ig- 
nited a long time at bright redness, there results a double sul- 
phide of aluminium and sodium, from which aluminium can be 
easily extracted. 

It has been stated § that if aluminium fluoride is heated with 

* English Patent, 1858, No. 461. 
t English Patent, No. 4,756, March 28, 1887. 
JPolytechnisches Central. Blatt., 1858, p. 888. 
§ Chemical News, i860. 


calcium sulphide, aluminium sulphide results. F. Lauterborn* 
also makes the same claim in a patent twenty years later, but 
the possibility of this reaction taking place is not yet beyond 

* German Patent (D. R. P.), No. 14,495 (1880). 



Some years ago, in order to treat fully of the metallurgy of 
aluminium it would have been as necessary to accompany it 
with all the details of the manufacture of sodium as to give the 
details of the reduction of the aluminium, because the manu- 
facture of the, former was carried on solely in connection with 
that of the latter. But now sodium has come out of the list of 
chemical curiosities, and has become an article of commerce, 
used for many other purposes than the reduction of aluminium. 
In fact, the last four years has seen the almost total extinction 
of the sodium processes of producing aluminium. If it be asked 
why such a chapter as this is still retained in a treatise on alu- 
minium, the answer is that for thirty years aluminium was 
made solely by its use; sodium is bound up with the history of 
aluminium ; and it is not impossible, though perhaps improba- 
ble, that some improved form of sodium process, in which me- 
tallic sodium is produced and at once utilized in the one ope- 
ration, may yet find a footing in the aluminium industry. For 
these and other reasons which might be advanced, the methods 
of producing sodium are yet of considerable interest to the 
worker in aluminium, and are therefore retained, with the addi- 
tion of a few recent processes. 

Davy to Deville (1808-1855). 

Sodium was first isolated by Davy by the use of electricity in 

the year 1808.* Later, Gay Lussac and Thenard made it by 

decomposing at a very high temperature a mixture of sodium 

carbonate and iron filings, f In 1808, also, Curaudau an- 

* Phil. Trans., :8o8. 
fRecherches Physico-chimiques, 1810. 



nounced that he had succeeded in producing potassium or 
sodium without using iron, simply by decomposing their car- 
bonates by means of animal charcoal. Briinner, continuing 
this investigation, used instead of animal charcoal the so-called 
black flux, the product obtained by calcining crude tartar from 
wine barrels. He was the first to use the wrought-iron mer- 
cury bottles. The mixture was heated white hot in a furnace, 
the sodium volatilized, and was condensed in an iron tube 
screwed into the top of the flask, which projected from the fur- 
nace and was cooled with water. In Brunner's experiments he 
only obtained three per cent, of the weight of the mixture as 
metallic sodium, the rest of the metal being lost as vapor. 
Donny and Mareska gave the condenser the form which with 
a few modifications it retains to-day. 
It was of iron, 4 millimetres thick, and 
was made in the shape of a book, 
having a length of about 100 centi- 
metres, breadth 50, and depth 6 (see 
Fig. 9). This form is now so well 
known that a further description is 
unnecessary. With this condenser the 
greatest difficulty of the process was 
removed, and the operation could be 
carried on in safety. This apparatus 
was devised and used by Donny and Mareska in 1854, with the 
supervision of Deville. 

FlQ. 9. 

Deville' s Improvements at Javel (1855). 

The following is Deville's own description of the attempts 
which he made to reduce the cost of producing sodium. As 
far as we can learn these experiments were commenced in 1854, 
but the processes about to be given are those which were car- 
ried out at Javel, March to June, 1855. As the description 
contains so many allusions to the difficulties met not only in pro- 
ducing but also in handling and preserving sodium, its perusal 
is yet of value to all concerned in this subject, although the 


actual methods here described have been superseded by much 
more economical ones. 

Properties of sodium. — "The small equivalent of sodium and 
the low price of sodium carbonate should long since have 
caused it to be preferred to potassium in chemical operations, 
but a false idea prevailed for a long time concerning the diffi- 
culties accompanying the reduction. When I commenced 
these researches the cost of sodium was at least double that of 
potassium. In this connection I can quote from my memoir 
published in the Ann. de Chim. et de Phys., Jan., i, 1855 : "I 
have studied with care the preparation of sodium and its prop- 
erties with respect to oxygen and the air, in order to solve the 
difficulties which accompany its reduction and the dangers of 
handling it. In this later respect, sodium is not to be com- 
pared to potassium. As an example of how dangerous the 
latter is, I will relate that being used to handle sodium and 
wishing once to replace it with potassium, the simple rubbing 
of the metal between two sheets of paper sufficed to ignite it 
with an explosion. Sodium may be beaten out between two 
sheets of paper, cut and handled in the air, without accident if 
the fingers and tools used are not wet. It may be heated with 
impunity in the air, even to its fusing point, without taking fire, 
and when melted, oxidation takes place slowly, and only at the 
expense of the moisture of the air. I have even concluded 
that the vapor alone of sodium is inflammable, but the vivid 
combustion of the metal can yet take place at a temperature 
which is far below its boiling point, but at which the tension of 
the metallic vapors has become sensible." I will add to these 
remarks that sodium possesses two considerable advantages : 
it is obtained pure at the first operation, and, thanks to a 
knack which I was a long time in finding out, the globules of 
the metal may be reunited and treated as an ordinary metal 
when melting and casting in the air. I have thus been able to 
dispense with the distillation of the raw products in the manu- 
facture — an operation which had come to be behaved neces- 
sary, and which occasioned a loss of 50 per cent, or so on the 


return, without appreciable advantage to the purity of the 
metal. The manufacture of sodium is in no manner encum- 
bered by the carburetted products, or perhaps nitrides, which 
are very explosive, and render the preparation of potassium so 
dangerous. I ought to say, however, that by making potas- 
sium on a large scale by the processes I am about to describe 
for sodium, Rousseau Bros, have diminished the dangers of its 
preparation very much, and practice the process daily in their 
chemical works. 

Method employed. The method of manufacture is founded 
on the reaction of carbon on alkaline carbonate. This method 
has been very rarely applied to sodium, but is used every day 
for producing potassium. Brunner's process is, in fact, very 
difficult to apply, great trouble being met, especially in the 
shape of condenser used. It is Donny and Mareska who have 
mastered the principles which should guide in constructing 
these condensers. 

Composition of mixtures used. The mixture which has given 
me excellent results in the laboratory is : 

Sodium carbonate 717 parts. 

Wood charcoal 1 75 " 

Chalk 108 " 


Dry carbonate of soda is used, the carbon and chalk pulver- 
ized, the whole made into a paste with oil and calcined in a 
crucible. The end of a mercury bottle, cut off, serves very 
well, and can be conveniently closed. Oil may be used alto- 
gether in place of charcoal, in which case the following pro- 
portions are used : 

Sodium carbonate 625 parts. 

Oil 280 " 

Chalk 95 " 

That a mixture be considered good, it should not melt at the 
temperature at which sodium is evolved, becoming liquid at 


this point and so obstructing the disengagement of the gas.* 
But it should become pasty, so as to mold itself evenly against 
the lower side of the iron vessel in which it is heated. The 
considerable latent heat required by carbonic oxide and sodium 
in assuming the gaseous state, is one cause of cooling which re- 
tards the combustion of the iron. When soda salt is intro- 
duced in place of dried soda crystals, the mixture, whatever its 
composition, always melts, the gases making a sort of ebulli- 
tion, the workmen saying that the apparatus " sputters." This 
behavior characterizes a bad mixture. It has been demon- 
strated to me that the economy made at the expense of a 
material such as carbonate of soda, the price of which varies 
with its strength in degrees, and which forms relatively a small 
portion of the cost of sodium, is annulled by a decrease of 20 to 
25 per cent, in the return of sodium. The oil used ought to be 
dry and of long flame. It acts as a reducing agent, and also 
furnishes during the whole operation hydrogenous gases, and 
even, towards the close, pure hydrogen, which help to carry 
the sodium vapor rapidly away into the condenser, and to 
protect the condensed metal from the destructive action of the 
carbonic oxide. Oil renders a similar service in the manu- 
facture of zinc. The role of the chalk is easy to understand. 
By its infusibility it decreases the liability of the mixture to 
melt. Further it gives off carbonic acid, immediately reduced 
by the carbon present to carbonic oxide. Now, the sodium 
ought to be carried rapidly away out of the apparatus, because 
it has the property of decomposing carbonic oxide, which is 
simultaneously formed, within certain limits of temperature, 
especially if the sodium is disseminated in little globules and 
so presents a large surface to the destructive action of the gas. 
It is necessary, then, that the metallic vapors should be rapidly 
conducted into the condenser and brought into the liquid state 

* It seems plain, however, granting that vapors would be evolved most freely from 
2. perfectly infusible charge, that a pasty condition, such as is recommended in the 
next sentence, would be the worst possible state of the charge for evolving gas, being 
manifestly inferior to a completely fluid bath.— J. W. R. 


— not into that state comparable to "flowers of sulphur," in 
which the metal is very oxidizable because of its fine division. 
A rapid current of gas, even of carbonic oxide, actively carries 
the vapors into the condenser, which they keep warm and so 
facilitate the reunion of the globules of sodium. At La 
Glaciere and Nanterre a mixture was used in which the, pro- 
portion of chalk, far from being diminished, was, on the con- 
trary, increased. The proportions used were — 

Sodium carbonate 40 kilos = 597 parts. 

Oil 18 " =269 " 

Chalk 9 " =134 " 

67 1000 

This quantity of mixture ought to give 9.4 kilos of sodium, 
melted and cast into ingots, without counting the metal divided 
and mixed with foreign materials, of which a good deal is 
formed. This return would be one-seventh of the weight of 
the mixture, or one-quarter of the sodium carbonate used. 

Use of these mixtures. The carbonate of soda, charcoal, and 
chalk ought to be pulverized and sieved, well mixed and again 
sieved, in order to make a very intimate mixture ; the mixture 
ought to be used as soon as possible after preparing, that it 
may not take up moisture. The mixture may be put just as it 
is into the apparatus where it should furnish sodium, but it may 
very advantageously be previously calcined so as to reduce its 
volume considerably, and so permit a greater weight being put 
into the same vessel. I believe that whenever this calcination 
may be made with economy, as with the waste heat of a fur- 
nace, a gain is made by doing so, but the procedure is not in- 
dispensable. However, the utility of it may be judged when 
it is stated that a mercury bottle held two kilos of non-calcined 
mixture, but 3.6 kilos were put into one when previously cal- 
cined. These two bottles heated in the same fire for the same 
time gave quantities of sodium very nearly proportional to the 
weight of soda in them. In working under the direction of a 
good workman, who made the bottles serve for almost four 


operations, I have been able to obtain very fine sodium at as 
low a price as 9.25 francs per kilo. ($0.84 per pound). In the 
manufacture of sodium by the continuous process, where the 
materials may be introduced red hot into the apparatus, this 
preliminary calcination is a very economical operation. 

Apparatus for reducing, condensing, and heating. — M. Briin- 
ner had the happy idea of employing mercury bottles in manu- 
facturing potassium ; thus the apparatus for reduction was in the 
hands of any chemist, and at such a low price that any one has 
been able to make potassium without much trouble. These 
bottles are equally suitable for preparing sodium, and the 
quantity which may be obtained from such apparatus and the 
ease with which they are heated, are such that they might have 
been used a long time for the industrial manufacture, except 
for two reasons which tend to increase the price of these bottles 
continually. For some time a large number of bottles have 
been sent to the gold workers of Australia and California, also 
large quantities have been used in late years in preparing the 
alkaline metals ; these two facts have diminished the number to 
such a point that from 0.5 or i franc the price has been raised 
to 2.5 or 3 francs. It has thus become necessary to replace 
them, which has been done by substituting large wrought-iron 
tubes which have the added advantage of being able to be 
worked continuously. I will first describe the manufacture in 
mercury bottles, which may still be very advantageously used 
in the laboratory, and afterwards the continuous production in 
large iron cylinders as now worked industrially. 

Manufacture in mercury bottles. — The apparatus needed is 
composed of a furnace, a mercury bottle, and a condenser. 

The form of furnace most suitable is a square shaft, C (Fig. 
10), the sides of which are refractory brick, while the grate G 
ought to have movable grate bars, the furnace being connected 
above with a chimney furnishing good draft. The flue F, con- 
necting with the chimney, should have a damper, R, closing 
tightly, and should lead exactly from, the centre of, the top of 
the shaft, thus dividing the draft equally all over the grate. 

1 86 


Coke is charged through lateral openings at 0. A small open- 
ing closed by a brick should be left a short distance above the 
grate bars, in order to poke down the coke around the bottle 
should it not fall freely. The space between the grate and 
bottle should always be full of fuel in order to keep the iron of 
the bottle from being burnt. In front of. the furnace is a square 
opening, P, closed with an iron plate, which has a hole in it by 
which the tube T issues from the furnace. 

The mercury bottle is supported on two refractory bricks, 
KK, cut on their top side to the curve of the bottle. These 

Fig. 10. 

should be at least 20 centimetres high to maintain between the 
grate and the bottle a convenient distance. The illustration 
gives the vertical dimensions correctly, but the horizontal 
dimensions are somewhat shortened. There should be at least 
12 centimetres between the bottle and sides of the furnace. 
However, all these dimensions should vary with the strength of 
the chimney draft and the kind of fuel used ; the furnace 
should be narrower if the draft is very strong and the coke 
dense. The iron tube T, which may conveniently be made of 
a gun barrel, is either screwed into the bottle or it may be 
simply fitted and forced into place, provided it hold tightly 


enough. It should be about 5 to 6 centimetres long, and 
should project scarcely i centimetre from the furnace. The 
end projecting should be tapered off in order to fit closely into 
the neck of the condenser. 

The condenser is constructed with very little deviation from 
that given by Donny and Mareska (see Fig. 9, p. 180). I 
have tried my best to make this apparatus as perfect as 
possible, but have always reverted to the form described by 
those authors ; yet even the very small differences I have made 
are indispensable and must be rigidly adhered to if it is wished 
to get the best results obtainable. Two plates of sheet iron, 2 
to 3 millimetres thick, are taken and cut into the shape indi- 
cated by Fig. II. One plate. A, remains flat except at the 
point C, where it is drawn by hammering into a semi- 
cylindrical neck of about 25 millimetres inside diameter. This 

Fig. II. 

iiiiii 111! 

■corresponds with a similar neck in the other plate, so that on 
joining the two there is a short cylinder formed. The edges of 
the plate A are raised all around the sides about 5 to 6 milli- 
metres, so that when the two plates are put together the longi- 
tudinal section from £> to Cis as in Fig. 12. As to the end, 
in one form the edge was not turned up, leaving the end open 
as in Fig. 13. Another form, which I use when wishing to let 
the sodium accumulate in the condenser till it is quite full, is 
made by turning up the edge at the end all but a small space 
left free, thus giving the end the appearance of Fig. 14, this 
-device also being shown in Fig. 11. The gas evolved during 


the reaction then escapes at the hole 0. The most rational 
arrangement of the apparatus is that shown in Fig. 15. In this 
condenser the lower part, instead of being horizontal, is in- 
clined, and the end having two openings, O and 0' , the sodium 
trickles out at the lower one as it condenses, while the gas es- 
capes by the slightly larger upper opening. In placing the 
plates together, the raised edges are washed with hme so as to 
form a good joint with the flat plate, and the plates are kept 
together by strong pressure grips. 

To conduct the operation, the bottles are filled entirely with 
mixture, the tube T adjusted, and then placed on the two sup- 
ports, there being already a good bed of fire on the grate. The 
front is put up, the shaft filled with coke, and the damper 
opened. The gases disengaged from the bottle are abundant, 
of a yellow color ; at the end of half an hour white fumes of 
carbonate of soda appear. The condenser should not yet be 
attached, but it should be noted if any sodium condenses on a 
cold iron rod pushed into the tube, which would be indicated 
by its fuming in the air. As soon as this test shows that sodium 
is being produced, the condenser is attached and the fire kept 
quite hot. The condenser soon becomes warm from the gases 
passing through it, while the sodium condenses and flows out 
at the end D (Fig. 15). It is received in a cast-iron basin L, 
in which some non-volatile petroleum is put. When at the end 
of a certain time the condenser becomes choked, it is replaced 
by another which has been previously warmed up to 200° or 
300° by placing it on top of the furnace. If the closed con- 
denser is used, care must be taken to watch when it becomes 
full, on the point of running from the upper opening, and the 
condenser then replaced and plunged into a cast-iron pot full 
of petroleum at a temperature of 150°. The sodium here 
melts at the bottom of this pot and is ladled out at the end of 
the day. The oil is generally kept up to 150° by the hot con- 
densers being plunged in constantly. The pot ought to have 
a close cover, to close it in case the oil takes fire ; the extinc- 
tion of the fire can thus be assured and no danger results. If 


the oil fires just as a condenser is being introduced, the sodium 
is run out in the air without igniting, the only drawback being 
that the condenser must be cleaned before using again. This 
method occasions a large loss of oil, however, and has been 
completely abandoned for the other form of condensers. When 
the operation proceeds well only pure sodium is obtained, the 
carbonized products which accompany in so provoking a man- 
ner the preparation of potassium not occurring in quantity 
sufficient to cause any trouble. Before using a condenser a 
second time it is put on a grating over a basin of petroleum 
and rubbed with a chisel-pointed tool in order to remove any 
such carbonized products. From time to time this material is 
■collected, put into a mercury bottle, and heated gently. The 
oil first distils, and is condensed in another cold bottle. The 
fire is then urged, a condenser attached, and the operation pro- 
ceeds as with a fresh charge, much sodium being thus recov- 

The raw sodium is obtained from the bottles in quantities 
of over half a kilo ; it is perfectly pure, dissolving in absolute 
alcohol without residue. It is melted and moulded into ingots 
just as lead or zinc. The operation I have described is exe- 
cuted daily, and only once has the sodium ignited. To prevent 
such accidents it is simply necessary to keep water away from 
the apparatus. The reduction of carbonate of soda and the 
production of sodium are such easy operations that when tried 
by those conversant with the manufacture of potassium or who 
have read about the difficulties of the production of sodium, 
success is only gained after several attempts — the failure being 
due solely to excess of precautions. The reduction should be 
carried on rapidly, so that a bottle charged with two kilos of 
rriixture may be heated and emptied in, at most, two hours. It 
is unnecessary to prolong the operation after the yellow flame 
stops issuing from the condenser, for no more sodium is ob- 
tained and the bottle may frequently be destroyed. The 
temperature necessary for the reduction is not so high as it has 
teen so far iiriagined. M. Rivot, who has assisted in these ex- 


periments, thinks that the bottles are not heated higher than 
the retorts in the middle of the zinc furnaces at Vielle Mon- 
tagne. I have been even induced to try cast-iron bottles, but 
they did not resist the first heating, without doubt because they 
were not protected from the fire by any luting or covering. But 
I was immediately successful in using cast-iron bottles decar- 
burized by the process used for making malleable castings. 
The mercury bottles heated without an envelope ought to serve 
three or four operations when entrusted to a careful workman. 
Besides all these precautions, success in this work depends 
particularly on the ability and experience of the workman, who 
can at any time double the cost of the sodium by carelessness 
in managing the fire. _ 

Continuous manufacture in cylinders. It might be thought 
that by increasing proportionately in all their parts the dimen- 
sions of the apparatus just described, it would be easy to pro- 
duce much larger quantities of sodium. This idea, which 
naturally presented itself to me at once, has been the cause of 
many unfruitful attempts, into the details of which I will not 
enter. I must, however, explain some details which may ap- 
pear insignificant at first sight, but which were necessitated 
during the development of the process. For instance, it will 
perhaps look irrational for me to keep the same sized outlet 
tubes and condensers that were used with the mercury bottles, 
for tubes five times as large ; but I was forced to adopt this 
arrangement after trying the use of tubes and condensers of all 
sizes ; indeed, it is fortunate for the success of the operation 
that this was so, for it became very injurious to the workmen 
to handle the large and weighty apparatus in the face of a large 
sodium flame. 

The mixture of sodium carbonate and carbon is made in the 
manner already described. I would say again that a previous 
strong calcination of the materials presents a great advantage^ 
not only because it permits putting a much larger weight into 
the retorts at once, but also that, being more compact, the mix- 
ture will not rise as powder and be violently thrown out of the 



Strongly-heated retorts. The mixture should also be calcined 
as needed, and used to fill the tubes while still red hot. When 
cold, uncalcined mixture is used, it is put into large cartridges 
of thick paper or canvas, 8 centimetres diameter and 35 cen- 
timetres long. 

The furnace and tubes are shown in section in Fig. 16. The 
tubes T are 120 centimetres long, 14 centimetres inside 
diameter, and 10 to 12 millimetres in thickness. They are 
formed from one piece of boiler iron, bent and welded along 

Fig. 16. 

one side. The iron plate P which closes one end is about 2 
centimetres thick, and pierced on one of its edges quite close 
to the side of the cylinder by a hole in which is screwed or 
fitted an iron tube, Z, 5 to 6 centimetres long and 15 to 20 
millimetres inside diameter, and tapering off at the end to fit 
into the condenser neck. The other end of the tube is closed 
by an iron plug, 0, terminated by a knob. The welded side of 
the tube is kept uppermost. These iron tubes should not, like 
the mercury bottles, be heated in the bare fire ; it is necessary 
to coat them with a resistant luting which is itself enveloped by 
a refractory jacket i centimetre thick, 22 centimetres interior 


diameter, and the same length as the retorts. This protection 
is commenced by plastering the retorts over with a mixture of 
equal parts of raw clay and stove ashes, which have been made 
into a paste with water and as much sand worked into the mix- 
ture as it will take without losing its plasticity, also adding 
some horse manure. This luting should be dried slowly, and 
the tube thus prepared is introduced into the refractory jacket, 
the open space between the two being filled with powdered 
refractory brick. Finally, luting is put on the iron plate P, so 
that no part of it is exposed to the flame. 

The furnace I have used is a reverberatory, but I do not re- 
commend its use without important modifications, because it 
does not realize all the conditions of easy and economic heat- 
ing. The grate is divided into two parts by a little wall of re- 
fractory brick, on which the middle of the reduction cylinders 
rests. The tubes are thus seen to be immediately over the bed 
of fuel. The top of the bridge is a little higher than the upper 
edge of the cylinders, this and the very low arch making the 
flame circulate better all around the tubes. A third cylinder 
might easily be placed above these two, and be heated satis- 
factorily, without any more fuel being burnt. This reverbera- 
tory receives on its bed the mixtures to be calcined, placed in 
cast-iron or earthen pots according to their composition. 
When the furnace is kept going night and day producing 
sodium, the temperature rises on the bed to clear cherry-red, 
and experience has shown that other reducing cylinders might 
be placed there, under such conditions, and be heated suf- 
ficiently for the reduction. 

All that I have said of the manufacture of sodium in mercury 
bottles applies equally to its manufacture in cylinders. The 
only difference consists in the charging and discharging, and I 
have only to add several precautions to be taken. On intro- 
ducing the cartridges containing uncalcined mixture, only 8 to 
9 kilos can be heated at once ; double as much can be used of 
previously-calcined mixture. The plug is put in place, not 
so tightly that it cannot easily be taken out again ; a little 


luting stops all leakages which show themselves. The reduc- 
tion lasts about four hours. When it is finished, a little water is 
thrown on the plug 0, and it is easily loosened and removed. 
On looking into the cylinder, the cartridges are seen to have 
kept their shape, but have shrunken so much that their diameter 
is only about 2 to 3 centimetres ; they are very spongy. This 
shows that the mixture has not melted ; the remainder is prin- 
cipally lime and carbon, and free from sodium carbonate. 
While opening the cylinder, a bright-red-hot iron is thrust into 
the outlet tube L, to keep dirt from getting into it, and it is 
kept in until the charging is finished. The cartridges are put 
in by means of semi-cylindrical shovels. The sudden heating 
of the mixture disengages soda dust from uncalcined mixtures, 
which is very disagreeable to the workmen. The cylinders are 
closed, and when the sodium flame appears at the outlet tube 
the condenser is attached, and the operation proceeds as 
already described. 

The envelopes of the cylinders are thick enough to prevent 
the distillation of the sodium being in any way affected by the 
accidental causes of cooling the fire. So when fresh fuel is 
charged or the door of the reverberatory is opened, causing the 
draft to cease almost entirely in the fire-place, the operation 
should not suffer by these intermittences, provided.that they are 
not too long prolonged. In short, when operating in cylinders, 
the production of sodium is easier, less injurious to the work- 
men, and less costly in regard to labor and fuel than when 
working with mercury bottles. At times, after working a fort- 
night with many interruptions dangerous for the apparatus, my 
experiment has been suddenly ended. The furnace was intact ; 
the envelopes of the tubes were split open, and the luting on 
the tubes found to be compact and coherent, but without traces 
of fusion, showing perfect resistance. The iron tubes mean- 
while had not suffered inside or out, and seemed as though 
they would last indefinitely. I attribute this success to the par- 
ticular care given to the composition of the jackets, and to the 
perfection with which the tubes had been welded. Only on 


one of the tubes was a very slight crack found, on a part not 
the most highly heated, and not sufficient to cause the tube to 
be discarded. 

Tissier Bros.' method of procedure. (1856). As related in 
the historical treatment of the subject (p. 13), Deville charged 
the Tissier Bros, with appropriating from him the process for 
the continuous production of sodium in cylinders, which, as 
just given, was devised during the experiments at Javel. On 
the other hand, the Tissier Bros, asserted their right to the 
process, patenting it, and using it in the works started at Rouen 
in the latter part of 1855. The following details are taken 
from Tissier's " Recherche de 1' Aluminium," only such being 
selected as supplement Deville's description, which has just 
been given. 

The sodium carbonate is first well dried at a high temper- 
ature, then mixed with well-dried pulverized charcoal and chalk, 
ground to the finest powder, the success of the operation de- 
pending on the fineness of this mixture. The proportions of 
these to use are various. One simple mixture is of 

Sodium carbonate 566 

Coal 244 

Chalk 95 

Coke 95 


Another contains — 

Sodium carbonate 615 

Coal 277 

Chalk 108 


The addition of chalk has the object of making the mixture 
less fusible and more porous, but has the disadvantage that the 
residue remaining in the retort after the operation is very im- 
pure, and it is impossible to add any of it to the succeeding 
charge ; and also, some of it being reduced to caustic lime 
forms caustic alkali with some sodium carbonate, which is then 



lost. WheD the mixture is well made it is subjected to a pre- 
limiinary calcination. This is done in cast-iron cylinders, two 
of which are placed side by side in a furnace and heated to 
redness (see Fig. 17). This is continued till all the moisture, 
carbonic acid, and any carburetted hydrogen from the coal, 
cease coming off. The mass contracts, becomes white and 
somewhat dense, so that a larger amount of the mixture can 
now be treated in the retorts where the sodium is evolved. As 
soon as the outcoming gases burn with a yellow flame, showing 

Fig. 17. 

sodium coming ofT, the calcination is stopped. The mixture is 
then immediately drawn out on to the stone floor of the shop, 
where it cools quickly and is then ready for the next operation. 
This calcination yields a mixture which without any previous 
reactions is just ready to evolve sodium when brought to the 
necessary temperature. This material is made into a sort of 
cylinder or cartridge and put into the decomposition retorts 
(see Fig. 16). The charging should be done quickly. The 
final retorts are of wrought-iron, since cast-iron would not 
stand the heat. At each end this retort is closed with wrought- 
iron stoppers and made tight with fire-clay. Through one 
stopper leads the pipe to the condenser, the other stopper is 
the one removed when the retort is to be recharged. These 
retorts are placed horizontally in rows in a furnace. Usually 
four are placed in a furnace, preferably heated by gas, such as 
the Siemens regenerative furnace or Bicheroux, these being 


much more economical. In spite of all these precautions the 
retorts will be strongly attacked, and in order to protect them 
from the destructive action of a white heat for seven or eight 
hours they are coated with some kind of fire-proof material. 
The best for this purpose is graphite, which is made into cylin- 
ders enclosing the retorts, and which can remain in place till 
the furnace is worn out. These graphite cylinders not only 
protect the iron retorts, but prevent the diffusion of the gas- 
eous products of the reaction into the hearth, and so support 
the retorts that their removal from the furnace is easily accom- 
plished. Instead of these graphite cylinders the retorts may 
be painted with a mixture that melts at white heat and so en- 
amels the outside. A mixture of alumina, sand, yellow earth, 
borax, and water-glass will serve very well in many cases. We 
would remark that the waste gases from this furnace can be 
used for the calcining of the mixture, or even for the reduction 
of the aluminium by sodium, where the manufacture of the 
former is connected with the making of the sodium. 

As for the reduction of the sodium, the retort is first heated 
to redness, during which the stopper at the condenser end of 
the retort is left off. The charge is then rapidly put in, and the 
stopper at once put in place. The reaction begins almost at 
once and the operation is soon under full headway, the gases 
evolved burning from the upper slit of the condenser tube with 
a flame a foot long. The gases increase in volume as the 
operation continues, the flame becoming yellower from sodium 
and so intensely bright as to be insupportable to look at. Now 
has come the moment when the workman must quickly adapt 
the condenser to the end of the tube projecting from the retort, 
the joint being greased with tallow or parafHn. The sodium 
collects in this in a melted state and trickles out. The length 
of the operation varies, depending on the intensity of the heat 
and the quantity of the mixture ; a charge may sometimes be 
driven over in two hours, and sometimes it takes eight. We 
can say, in general, that if the reaction goes on quickly a some- 
what larger amount of sodium is obtained. The higher the 


heat used, however, the quicker the retorts are destroyed. The 
operation requires continual attention. From time to time, a 
workm-an with a prod opens up the neck of the condenser, 
but if care is not taken the metal overflows ; if this happens, 
the metal overflowing is thrown into some petroleum, while 
another man replaces the condenser with an empty one. The 
operation is ended when the evolution of gas ceases and the 
flame becomes short and feeble, while the connecting tube be- 
tween the retort and condenser keeps clean and does not stop 
up. As soon as this occurs, the stopper at the charging end 
is removed, the charge raked out into an iron car, and a new 
charge being put in, the operation continues. After several 
operations the retorts must be well cleaned and scraped out. 
The sodium thus obtained is in melted bits or drops, mixed 
with carbon and sodium carbonate. It must, therefore, be 
cleaned, which is done by melting it in a wrought-iron kettle 
under paraffin with a gentle heat, and then casting it into the 
desired shapes. The sodium is kept under a layer of oil or 
any hydrocarbon of high boiling point containing no oxygen. 
Tissier gives the reaction as — 


The sodium is condensed, while the carbonic oxide, carrying 
over some sodium, burns at the end of the apparatus. This 
would all be very simple if the reaction of carbonic oxide on 
sodium near the condensing point did not complicate matters, 
producing a black, infusible deposit of sodium monoxide 
(Na^O) and carbon, which on being melted always give rise to 
a loss of sodium. 

Deville's Improvements at La Glacier e (1857). 

At this works Deville tried the continuous process of manu- 
facturing sodium in cylinders on a still larger scale, with the fol- 
lowing results, as described by Deville himself: — 

" We made no change in the composition of the mixtures used 


from those already described, or in the form or size of the iron 
tubes or the method of condensation ; but we worked with six 
cylinders at a time in a furnace similar to the puddling furnaces 
of M. Guadillot, the tubes being protected by refractory envel- 
opes. The cylinders were so arranged on the hearth that the 
flame bathed all parts of their surface. A low brick wall ex- 
tends down the centre of the hearth, supporting the middle of the 
cylinders, which extend across it. The hearth is well rammed 
with refractory sand, and the space between it and the bottom 
of the cylinder serves as a passage-way for most of the flame. 

" Our six cylinders worked satisfactorily for five days. We 
were able to observe that they were all heated with remarkable 
uniformity, and that the heat was sufficient all round them. It 
also appeared that the rear end of the cylinders required only 
a hermetic seal. Indeed, as soon as the operation was well 
under way and sodium distilling off, some of it condensed and 
oxidized in the cool parts of the apparatus, forming a sort of 
plug of carbonate and carbides of sodium, which the vapor and 
gases could no longer penetrate. We were thus able for a 
long time to distil sodium away from one of our tubes which 
was entirely opened at the rear. 

" This new furnace worked so well that we were hopeful of 
complete success, when an accident happened which compelled 
the stopping of the experiment. The iron tubes had been 
ordered 1.20 metres long, the size of the hearth calculated ac- 
cordingly, but they were delivered to us only 1.05 metres long. 
We made use of these, with the result that the rear ends be- 
came red-hot during the operation and allowed sodium vapors 
to leak through. These leaked through the luting, and escap- 
ing into the furnace, melted the envelopes very rapidly. 

"In another attempt, in which this fault was avoided, we were 
unsuccessful because the envelopes gave way at the first heat- 
ing up, both they and the iron tubes being of inferior quality. 
We were considerably inconvenienced by the failure of these 
experiments, which caused considerable expense and gave no 
very definite results. Just then a new sort of apparatus was 


devised, a description of which is given later on. It will be 
seen that we were compelled to employ tubes of very small 
value, so that their destruction in case of accident involved no 
great loss, and to heat each one by an independent fire, so that 
the stoppage or destruction of one cylinder would not necessitate 
the stoppage or endanger the safety of the neighboring ones." 

Cast-iron vessels. Deville tried at La Glaciere, as well as at 
Javel, to utilize cast-iron vessels for producing sodium. Deville 
states the difficulties which caused their use to be unsuccessful 
to be as follows : 

"The result was always unfavorable. Sodium is obtained, but 
as soon as its production becomes rapid the vessel melts and 
the operation is quickly ended. This follows because the tem- 
perature necessary for the production of the metal is far from 
being sufficient for producing it in large quantities at once ; and 
we know that this is the one condition for condensing the so- 
dium well and obtaining it economically. This observation led 
me to think that by diminishing very much the temperature of 
the furnace, large apparatus of cast-iron with large working 
surface could be used, thus making at a time a large amount of 
metallic vapor which could be condensed in recipients of ordi- 
nary size. The whole large apparatus would thus have the out- 
put of a smaller one worked at a higher temperature. My ex- 
perience has shown me that in large-sized tubes heated to a 
low temperature there is formed in a given time about as much 
sodium as from a single mercury bottle at a much higher heat. 
This is the reason why larger condensers are not necessary 
with the larger tubes. Before knowing this fact, I tried a large 
number of useless experiments to determine the size of con- 
densers suitable for large apparatus. It is on this principle 
that I have long been endeavoring to make sodium without 
working at high temperatures, and using less costly and more 
easily protected apparatus." 

Improvements used at Nanterre (1859). 
The method used here was exactly that already described, 


the improvements being solely in details of the apparatus. 
These are described by Deville as follows : — 

" The experiments made at Javel and the continuous process 
used at Glaciere have shown us in the clearest manner the ab- 
solute necessity of efficient protection for the iron cylinders, for 
without this protection the method cannot be practiced with 
economy. Further, experiments in this direction are very 
costly, for the failure of a tube stops the working of a large 
number of cylinders, and often compromises the brick-work of 
the furnace itself. We therefore came to the conclusion that 
for making the small quantity of sodium we required, 300 to 500 
kilos a month, it would be better to employ smaller apparatus, 
independent of each other and easy to replace. 

" The iron tubes are made of thinner iron and at very little ex- 
pense, by taking a sheet of iron, curving it into a cylinder and 
riveting the seam. This tube resembles very closely those 
used at Javel, shown in Fig. 16, but of smaller dimensions. It 
is closed at each end by cast-iron plugs, one of which has a 
hole for the outlet tube. These cylinders are filled with sodium 
mixture and placed in furnaces of the form of Fig. 10, except it 
is necessary to have openings in the back and front of the fur- 
nace so that the cast-iron plugs closing the cylinders may be 
outside, to prevent their melting. We used coke at first for 
fuel, fed around the cylinders, but M. Morin has since placed the 
tubes out of direct contact with the fuel, uses soft coal, and heats 
the tubes by contact with the flame and by radiation. In the 
latest form used, two cylinders are placed in each furnace, and, 
in general, they serve for two or three operations. All that has 
been said in connection with the manufacture in mercury bot- 
tles is immediately applicable to the manufacture in cylinders 
of this kind, the capacity of which may vary from two to six or 
eight litres, without any change in the manner of using them. 
We have, however, adopted altogether condensers of cast-iron. 
The neck is cylindrical and belongs only to one-half of the ap- 
paratus, the neck end of the other plate being beveled and fit- 
ting closely against a recess in the other plate." 


The foregoing shows the sodium industry as it was perfected 
by Deville, in 1859, and as it remained for twenty-five years 
without sensible change. The cost of sodium by this process 
is stated to have been, in 1872, as follows: — 

Manufacture of one kilo of sodium. 

Soda 9.35 kilos @ 32 fr. per 100 kilos = 3 fr. 9 cent. 

Coal 74.32 " " 1.40" " " " = I " 4 " 

Wages 3 " 73 " 

Expenses 3 " 46 " 

Total 1 1 fr. 32 cent. 

which is equal to $1 per lb. The larger part of the expense 
account is the cost of retorts or tubes in which the operation 
takes place, and which are so quickly destroyed that the re- 
placing of them forms nearly one-quarter of the cost of the 

Minor Improvements (185 9- 1888). 

An experiment made by Deville in 1864* is of considerable 
interest in connection with some more recent processes. Guy 
Lussac and Thenard having observed that caustic potash is re- 
duced by iron, Deville took a mercury bottle with an aperture 
above and one below, to which was fitted an iron tube and a 
sodium condenser. Finely divided iron, reduced by hydrogen, 
was put into the bottle, and when red hot, caustic soda was in- 
troduced through the upper aperture. The iron was strongly 
acted upon, and in less than 20 minutes Deville obtained y^ a 
kilo of pure sodium. The reduction was stopped by a mixture 
of soda and iron oxide filling up the lower opening. The ar- 
rangement and reactions suggest strongly some later processes. 
Why Deville did not continue on this line, I cannot say. 

R. Wagnerf uses paraffin in preference to paraffin oil in 
which to keep the sodium after making it. Only pure paraffin, 
which has been melted a long time on a water-bath, and all its 
water driven off, can be used. The sodium to be preserved is 

* Lef ons de Chemie, 1864-5, P- 336- + Dingier, 1883, p. 252. 


dipped in the paraffin melted on a water-bath, and kept at no 
higher heat than 55° C, and the metal is thereby covered with a 
thick coat of paraffin which protects it from oxidation, and may 
then be put up in wooden or paper boxes. When the metal is 
to be used, it is easily freed from paraffin by simply warming 
it, since sodium melts at 95° to 96° C, and the paraffin at 50° 
to 60°. 

The reduction of potassium carbonate by carbon requires a 
much less degree of heat than that of sodium carbonate, and 
therefore many attempts have been made to reduce potassium 
and sodium together, under circumstances where sodium alone 
would not be reduced. Dumas* added some potassium car- 
bonate to the regular sodium mixture; and separated the 
sodium and potassium from each other by a slow, tedious oxi- 
dation. R. Wagnerf made a similar attempt. He says that 
not only does the reduction of both metals from a mixture of 
their carbonates with carbon work easier than sodium carbonate 
alone with carbon, but even caustic soda may be used with 
potassium carbonate and carbon. Also, the melting point of 
potassium and sodium alloyed is much lower than that of either 
one alone, in consequence of which their boiling point and the 
temperature required for reduction are lower. 

J. B. Thompson and W. WhiteJ specify mixing dry sodium 
carbonate with a liquid carbonaceous material, preferably tar, 
driving off all volatile matter in iron pots at a low heat, and 
then distilling in a tubular fire-clay retort connected with a 
tightly-closed receiver containing a little paraffin oil to ensure a 
non-oxiding atmosphere, and also provided with a small escape 
pipe for carbonic oxide. This process gave great prospects of 
success when tried in the laboratory, but on a manufacturing 
scale it failed for the reason (assigned by Mr. Thompson) that 
the sheet-iron tray, designed to keep the material from attack- 
ing the retort, absorbed carbon at about 1000° C. and fused, 
after which no sodium was produced, since the material took 

* Handbuch der Angewandten Chemie, 1830, ii. 345. 

t Dingier, 143, 343- J English Patent 8426, June 11, 1887. 


up silica from the retort, absorbing so much that the carbon no 
longer decomposed it. 

H. S. Blackmore,* of Mount Vernon, U. S. A., patents the 
following process of obtaining sodium: — 

Calcium hydrate 273^ parts. 

B erric oxide 31 " 

Dry sodium carbonate 31 " 

Charcoal loj!^ " 

are intimately mixed and subjected to a red heat for 20 min- 
utes, afterwards to a white heat. Caustic soda is first produced, 
the carbon reduces the ferric oxide, producing iron, which in 
its turn reduces the caustic soda, and sodium vapors distil. 
The residue consists of ferric oxide and lime, and is slaked and 
used over. 

O. M. Thowlessf of Newark, N. J., claims to place a retort in 
a furnace, providing it on one side with an arm through which 
carboniferous material can be supplied, on the other side with a 
similar arm (surrounded by flues), into which caustic soda or 
sodium carbonate is charged — a valve controlling their flow into 
the retort. Outside the furnace and on top of it is a flat con- 
denser into which the sodium vapor passes. 

G. A. JarvisI patents the replacement of the iron tubes or 
crucibles used in the manufacture of sodium, by fire-clay ap- 
paratus lined with basic material, such as strongly burnt mag- 
nesia with 10 per cent, of fluorspar. 

Castner's Process (1886). 
The first public announcement of this process was through 
one of the New York daily journals,^ and as the tone of the 
article is above that of the usual newspaper reports, and the 
expectations contained in it were subsequently more than 
realized, we cannot better introduce a description of this process 
than by quoting the paragraph referred to : — 

♦English Patent 15 156, Oct. 22, 1888. f English Patent 12486 (1887). 

X English Patent 4842, March 31, 1888. § New York World, May 16, 1886. 


"When sodium was reduced in cost to $1.50 per lb. it was 
thought to have touched a bottom figure, and all hope of mak- 
ing it any cheaper seemed fruitless. This cheapening was not 
brought about by any improved or new process of reduction, 
but was owing simply to the fact that the aluminium industry 
required sodium, and by making it in large quantities its cost 
does not exceed the above-mentioned price. The retail price 
is now $4.00 per lb. The process now used was invented by 
Briinner, in 1808, and up to the present time nothing new or 
original has been patented except three or four modifications of 
his process which have been adopted to meet the requirements 
of using it on a large scale. Mr. H. Y. Castner, whose labora- 
tory is at 218 West Twentieth Street, New York, has the first 
patent ever granted on this subject in the United States, and 
the only one taken out in the world since 1808. Owing to 
negotiations being carried on, Mr. Castner having filed appli- 
cations for patents in various foreign countries, but not having 
the patents granted there yet, we are not at liberty to state his 
process fully. The metal is reduced and distilled in large iron 
crucibles, which are raised automatically through apertures in 
the bottom of the furnace, where they remain until the reduc- 
tion is completed and the sodium distilled. Then the crucible 
is lowered, a new one containing a fresh charge is substituted 
and raised into the furnace, while the one just used is cleaned 
and made ready for use again. The temperature required is 
very moderate, the sodium distillling as easy as zinc does when 
being reduced. Whereas by previous processes only one-third 
of the sodium in the charge is obtained, Mr. Castner gets 
nearly all, for the pots are nearly entirely empty when with- 
drawn from the furnace. Thus the great items of saving are 
two or three times as much metal extracted from a given 
amount of salt, and cheap cast iron crucibles used instead of 
expensive wrought-iron retorts. Mr. Castner expects to pro- 
duce sodium at 25 cents per lb., thus solving the problem of 
cheap aluminium, and with it magnesium, silicon, and boron, 
all of which depend on sodium for their manufacture. Thus 


the production of cheap sodium means much more than cheap 
aluminium. Mr. Castner is well known in New York as a 
chemist of good standing, and has associated with him Mr. J. 
H. Booth and Mr. Henry Booth, both well known as gentlemen 
of means and integrity." 

The following are the claims which Mr. Castner makes in 
his patent:* 

1. In a process for manufacturing potassium or sodium, per- 
forming the reduction by diffusing carbon in a body of alkali 
in a state of fusion at moderate temperatures. 

2. Performing the reduction by means of the carbide of. a 
metal or its equivalent. 

3. Mechanically combining a metal and carbon to increase 
the weight of the reducing material, and then mixing this pro- 
duct with the alkali and fusing the latter, whereby the reducing 
material is held in suspension throughout the mass of fused 

4. Performing the deoxidation by the carbide of a metal or 
its equivalent. 

For an explanation of the principles made use of in the 
above outlined process we will quote from a lecture delivered 
by Mr. Castner at the Franklin Institute, Philadelphia, October 
1 2th, 1886. That Institution has since bestowed on Mr. Cast- 
ner one of its gold medals as a recognition of the benefit to 
science accruing from his invention. 

" In the ordinary sodium process, lime is added to the reduc- 
ing mixture to make the mass refractory, otherwise the alkali 
would fuse when the charge is highly heated, and separate from 
the light, infusible carbon. The carbon must be in the propor- 
tion to the sodium carbonate as four is to nine, as is found 
needful in practice, so as to assure each particle of soda in the 
refractory charge having an excess of carbon directly adjacent 
or in actual contact. Notwithstanding the well-known fact that 
sodium is reduced from its oxide at a degree of heat but slightly 
exceeding the reducing point of zinc oxide, the heat necessary 

* U. S. Pat. No. 342,897, June i, 1886. Hamilton Y. Castner, New York, 


to accomplish reduction by this process and to obtain even 
one-third of the metal in the chai-ge, closely approaches the 
melting point of wrought-iron. 

" In my process, the reducing substance, owing to its com- 
position and gravity, remains below the surface of the molten 
salt, and is, therefore, in direct contact with fused alkali. 
The metallic coke of iron and carbon contains about 30 per 
cent, carbon and 70 per cent, iron, equivalent to the formula 
FeCj. I prefer to use caustic soda, on account of its fusibility, 
and mix with it such quantity of so-called ' carbide ' that the 
carbon contained in the mixture shall not be in excess of the 
amount theoretically required by the following reaction: — 

3NaOHf FeC,=--3Na4-Fe + CO f CO.+sH ; 

or, to every 100 lbs. of pure caustic soda, 75 lbs. of 'carbide,' 
containing about 22 lbs. of carbon. 

" The necessary cover for the crucible is fixed stationary in 
each chamber, and from this cover a tube projects into the con- 
denser outside the furnace. The edges of the cover are convex, 
those of the crucible concave, so that when the crucible is raised 
into position and held there, the tight joint thus made prevents 
all leaking of gas or vapor. Gas is used as fuel, and the re- 
duction begins towards 1000° C. As the dharge is fused, the 
alkali and reducing material are in direct contact, and this fact, 
together with the aid rendered the carbon by the fine iron, in 
withdrawing oxygen from the soda, explains why the reduction is 
accomplished at a moderate temperature. Furthermore, by 
reducing from a fused mass, in which the reducing agent re- 
mains in suspension, the operation can be carried on in crucibles 
of large diameter, the reduction taking place at the edges of the 
mass, where the heat is greatest, the charge flowing thereto from 
the centre to take the place of that reduced. 

" I am enabled to obtain fully 90 per cent, of the metal in the 
charge, instead of 30 per cent, as formerly. The crucibles, 
after treatment, contain a little carbonate of soda, and all the 
iron of the 'carbide' still in a fine state of division, together 


with a small percentage of carbon. These residues are treated 
with warm water, the solution evaporated to recover the car- 
bonate of soda, while the fine iron is dried, and used over again 
for 'carbide.' " 

Mr. Castner having demonstrated in his New York laboratory 
the success of his process, went to England, and for several 
months during the winter of 1886-7 was engaged in building 
and working a large sodium furnace. This was successfully car- 
ried out near London, the inventor being assisted by Mr. J. 
MacTear, F. C. S., who, in March, 1887, read a description of 
this furnace and the results obtained before the Society of 
Chemical Industry. During the working of this furnace it was 
inspected by many chemical and metallurgical authorities, who 
were completely satisfied as to its success. As the furnace 
then described differed in a few details from the one just re- 
ferred to, it may be well to extract the essential particulars from 
Mr. MacTear's paper — on the ground that the importance of 
this invention justifies a complete discussion of its develop- 
ment: — 

" Since Mr. Castner's paper upon his process, which was 
read before the Franklin Institute of Philadelphia, October 12, 
1886, several slight changes in the mode of carrying on this 
process have been made. These have been brought about by 
the experience gained from the actual working of the process 
upon a commercially large scale. 

"The reactions by which the sodium is produced are some- 
what difficult to describe, as they vary somewhat according to 
the mixture of materials and temperature employed in the re- 
duction. The mixture and temperature which it is now pre- 
ferred to use is represented by the reaction : — 

6NaH0 + FeC2= 2Na2CO, + 6H + Fe + zNa. 

" In place of using an actual chemical compound of iron and 
carbon, as expressed by the above reaction, a substitute or 
equivalent is prepared as follows : To a given quantity of melted 
pitch is added a definite proportion of iron in a fine state of 


division. The mixture is cooled, broken up into lumps, and 
cooked in large crucibles, giving a metallic coke consisting of 
carbon and iron, the proportions of each depending upon the 
relative quantities of pitch and iron used. This metallic coke, 
after being finely ground, provides a substance having the iron 
and carbon in a like proportion to an iron carbide, and from 
which neither the iron nor carbon can be separated by mechan- 
ical means. The fine iron is conveniently prepared by passing 
carbonic oxide and hydrogen in a heated state, as obtained 
from an ordinary gas producer, over a mass of oxide of iron 
commercially known as 'purple ore,' heated to a temperature 
of about 500° C. 

" In producing sodium, caustic soda of the highest obtain- 
able strength is used, and there is mixed with it a weighed 
quantity of the so-called ' carbide,' sufficient to furnish the 
proper amount of carbon to carry out the reaction indicated 
above. The crucibles in which this mixture is treated are made 
of cast-steel, and are capable of containing a charge of 15 lbs. 
of caustic soda, together with the proper proportion of the 
• carbide.' 

"After charging a crucible with the above mixture, it is 
placed in a small furnace where it is kept at a low heat for 
about thirty minutes, during which time the mass fuses, boils 
violently, and a large part of the hydrogen is expelled by the 
combined action of the iron and carbon, the ' carbide,' owing 
to its gravity, remaining in suspension throughout the fused 
soda. At the end of the time stated, the contents of the cru- 
cible have subsided to a quiet fusion. The crucible is then 
lifted by a pair of tongs on wheels and placed upon the plat- 
form of the elevating gear, as shown in the drawing, (Fig. 18) 
and raised to its position in the heating chamber of the main dis- 
tilling furnace. The cover which remains stationary in the fur- 
nace has a convex edge, while the crucible has a groove round the 
edge into which the edge of the cover fits. A little powdered 
lime is placed in the crucible groove just before it is raised, so 
that when the edges of the cover and crucible come together 



they form a tight joint, and at the same time will allow the 
crucible to be lowered easily from the chamber when the ope- 
ration is finished, to give place to another containing a fresh 
charge. From the cover projects a slanting tube (see Fig. 18), 
connected with the condenser. The condenser is provided 
with a small opening at the further end to allow the escape of 
hydrogen, and has also a rod fixed (as shown), by means of 
which any obstruction which may form in the tube during dis- 

FlG. 18. 

tillation may be removed. After raising a crucible in its place 
in the furnace, the hydrogen escaping from the condenser is 
lighted, and serves to show by the size of the flame how the 
operation is progressing in the crucible, the sodium actually 
distilling soon after the crucible is in its place. The temper- 
ature of the reduction and distillation has been found to be 
about 823° C. The gas coming off during the first part of the 
distillation has been analyzed and found to consist of pure 
hydrogen. Analysis of the gas disengaged when the ope- 


ration was almost completed, gave as a result, hydrogen 95 per 
cent., carbonic oxide 5 per cent. It has been found advisable 
to use a little more 'carbide' than the reaction absolutely re- 
quires, and this accounts for the presence of the small quantity 
of carbonic oxide in the expelled gas, the free carbon acting 
upon the carbonate formed by the reaction, thus giving off car- 
bonic oxide and leaving a very small percentage of the residue 
in the form of peroxide of sodium. This small amount of car- 
bonic oxide rarely combines with any of the sodium in the 
tube, and so the metal obtained in the condensers is pure, and 
the tubes never become choked with the black compound. In 
the preparation of potassium a little less ' carbide' is used than 
the reaction requires ; thus no carbonic oxide is given off, and 
all danger attached to the making of potassium is removed. 
After the reduction and distillation the crucible is lowered from 
the furnace and the contents poured out, leaving the crucible 
ready to be recharged. The average analyses of the residues 
show their composition to be as follows: — 

Carbonate of soda 77 per cent. 

Peroxide of sodium 2 " 

Carbon 2 " 

Iron 19 " 

" The average weight of these residues from operating upon 
charges of 15 lbs. caustic soda and 5^ lbs. of carbide is 16 
lbs. These residues are treated either to produce pure crystal 
lized carbonate of soda or caustic soda, and the iron is recov- 
ered and used again with pitch in the formation of the ' car- 
bide.' From this residue weighing 16 lbs., is obtained 13 lbs. 
of anhydrous carbonate of soda, equivalent to 9.4 lbs. caustic 
soda of 76 per cent. 

" Operating upon charges as above mentioned the yield- has 
been — 

Sodium, actual 2.50 lbs. Theory 2.85 lbs. 

Soda carbonate, actual 13.00 lbs. " 13.25 lbs. 

" The average time of distillation in the large furnace has 


been i hour 30 minutes, and as the furnace is arranged for 
three crucibles, 45 lbs. of caustic soda are treated every 90 
minutes, producing 7^ lbs. of sodium and 39 lbs. of carbonate 
of soda. The furnace is capable of treating 720 lbs. of caustic 
soda daily, giving a yield in 24 hours of 120 lbs. of sodium and 
624 lbs. of anhydrous carbonate of soda. The furnace is heated 
by gas which is supplied by a Wilson Gas Producer, consuming 
I cwt. of fuel per hour. The small furnace in which the 
crucibles are first heated requires about J^ cwt. per hour. The 
following estimate of cost, etc., is given from the actual running 
of the furnace working with the above charges for 24 hours : — 

£ ^. d. 

720 lbs. of caustic soda @ ;^i i per ton 3 10 10 

1 50 lbs. of " carbide " @ yd^d. per lb o 6 4 

Labor I o o 

Fuel o 17 o 

Re-converting 624 lbs. of carbonate into caustic, at a cost 

of about £^ per ton on the caustic produced, say ... I o o 

Total 6 14 2 

Deducting value of 475 lbs. of caustic recovered 2 6 8 

Cost of 120 lbs. of sodium ^4 7 6 

Cost per pound, 83^</. 

" Regarding the item of cost relating to the damage caused 
to the crucibles by the heat, this question has been very care- 
fully gone into, some of the crucibles have been used upwards 
of fifty times, and from present indications of their condition 
there is no doubt that they can continue to be used at least 150 
times more before they become unfit for further use. In con- 
sidering 200 operations to be the life of a crucible, the item of 
damage or wear and tear amounts to less than \d. per lb. on 
the sodium produced; and if we take the furnace tear and wear 
at the same rate of \d. per lb., we will see that the tear and 
wear of plant is only one-twelfth of that incurred in the ordi- 
nary propess. It is upon these facts that Mr. Castner bases 
his claim to be able to produce sodium by Jiis- process upon 
the large scale, at a cost of less than 15. per lb. The advan- 


tages of this process will be apparent to any one at all familiar 
with the manufacture of these metals as conducted heretofore. 
The first and most important end gained is their cheap produc- 
tion, and this is owing chiefly to the low heat at which the 
metals are produced, the quickness of the operation, non- clog- 
ging of the conveying tubes, and a very small waste of mate- 
rials. The process furthermore admits of being carried on 
upon a very large scale ; in fact, it is intended ultimately to in- 
crease the size of the crucible so as to make the charges con- 
sist of 50 lbs. of caustic soda. Crucibles of cast iron have 
been found quite suitable, and it is intended in future, to use 
crucibles made of this material in place of the more expensive 

Immediately on the demonstration of this success, a company 
was formed to unite Mr. Castner's sodium process with Mr. 
Webster's improvements in the production of aluminium chlor- 
ide. The Aluminium Co., Ltd., first appeared before the public 
in June, 1887, and at the first meeting in the following Septem- 
ber it was decided to build works at once. These were begun 
at Oldbury, near Birmingham, and were in working operation 
by the end of July, 1888. The furnaces here erected were 
larger than the one just described, and altogether had a pro- 
ducing capacity of nearly a ton of sodium a day. The follow- 
ing details respecting this plant and its working are taken 
mostly from an address delivered before the Society of Arts, 
March 13, 1889, by Mr. William Anderson, and from a dis- 
course at the Royal Institution, May 3, 1889, by Sir Henry 
Roscoe, president of the company. 

There are four large sodium furnaces, each holding five pots 
or crucibles, and heated by gas, applied on the regenerative 
principle. A platform about five feet above the floor allows 
the workmen to attend to the condensers, while the lifts on 
which the pots are placed sink level with the floor. The 
crucibles used are egg-shaped, about 18 inches diameter at 
their widest part and 24 inches high ; when joined to the cover 
the whole apparatus is about 3 feet in height. The covers 


have vertical pipes passing through the top of the furnace, 
forming a passage for the introduction of part of the charge, 
and also a lateral pipe connecting with the condenser. The 
whole cover is fixed immovably to the roof of the furnace and 
is protected by brickwork from extreme heat ; but it can easily 
be removed when necessary. The natural expansion of the 
vessels is accommodated by the water pressure in the hydraulic 
lifts on which the pots stand. When the lift is lowered and 
sinks with the lower part of the crucible to the floor level, a 
large pair of tongs mounted on wheels is run up, and catching 

Fig. 19, 


hold of the crucible by two projections on its sides, it is carried 
away by two men to the dumping pits, on the edge of which it 
is turned on its side, the liquid carbonate of soda and finely- 
divided iron which form the residue are turned out, and the 
inside is scraped clean from the opposite side of the pit, under 
the protection of iron .shields. When clean inside and out, it 
is lifted again by the truck and carried back to the furnace, re- 
ceiving a fresh charge on its way. It is then put on the plat- 
form and lifted into place, having still retained a good red heat. 
It takes only i)^ to 2 minutes to remove and empty a crucible. 


and only 6 to 8 minutes to draw, empty, recharge, and replace 
the five crucibles in each furnace. The time occupied in re- 
ducing a charge is one hour and ten minutes. It is thus seen 
that one bank of crucibles yields 500 pounds of sodium in 
twenty- four hours, the battery of four furnaces producing about 
a ton in that time. 

The shape of the condenser has been altogether changed. 
Instead of the flat form used on the furnace at London (see 
Fig. 18), which resembled the condenser used in the Deville 
process, a peculiar pattern is used which is quite different. It 
consists in a tube-shaped cast-iron vessel S inches in diameter, 
nearly 3 feet long over all, and having a slight bend upwards at 
a point about 20 inches from the end. At this bend is a small 
opening in the bottom, which can be kept closed by a rod drop- 
ping into it; this rod, passing through a tight- fitting hole 
above, can be raised or lowered from outside. Thus the 
sodium can either run out continually into small pots placed 
beneath the opening or can be allowed to collect in the con- 
denser until several pounds are present, then a small potful run 
out at once, by simply lifting the iron rod. The outer end of 
the condenser is provided with a lid, hinged above, which can 
be thrown back out of the way when required. This lid also 
contains a small peep-hole covered with mica. In the top of 
the condenser just before the end is a small hole through 
which the hydrogen aud carbonic oxide gases escape when the 
end is closed, burning with the yellow sodium flame. The 
bend in the condenser is not acute enough to prevent a bar be- 
ing thrust through the end right into the outlet tube projecting 
from the furnace, thus allowing the whole passage to be cleaned 
out should it become choked up. Previous to drawing the 
crucibles from the furnace for the purpose of emptying them 
and recharging, the small pots containing the metal distilled 
from one charge are removed and empty ones put in their 
place. Those removed each contain on an average about 6 lbs. 
of sodium, or 30 lbs. from the whole furnace. When sufficiently 
cool, petroleum is poured on top of the metal in the pots, and 


they are -vvheeled on a truck to the sodium casting shop, where 
the sodium is melted in large pots heated by oil baths and cast 
either into large bars ready to be used for making aluminium 
or into smaller sticks to be sold. The sodium is preserved 
under an oil such as petroleum, which does not contain oxygen 
in its composition, and the greatest, care is taken to protect it 
from water. 

Special care is taken to keep the temperature of the furnace 
at about iOOO° C, and the gas and air-valves are carefully re- 
gulated so as to maintain as even a temperature as possible. 
The covers remain in the furnace from Sunday night to Satur- 
day afternoon, and the crucibles are kept in use till worn out, 
when new ones, previously heated red-hot, are substituted 
without interrupting the general running of the furnace. These 
bottom halves of the crucibles are the only part of the plant 
liable to exceptional wear and tear, and their durability is found 
to depend very much on the soundness of the casting, because 
any pores or defects are rapidly eaten into and the pot de- 
stroyed. The average duration of each crucible is now 750 
lbs. of sodium, or 125 charges. 

Apropos of the reaction involved in the reduction, it has 
probably been observed that Mr. MacTear proposes a different 
formula from that suggested by Mr. Castner. Mr. Weldon re- 
marked that when a mixture of sodium carbonate and carbon 
was heated the carbon did not directly reduce the soda, but at 
a high temperature the mixture gives off vapors of oxide of 
sodium (Na^O) part of which dissociates into free oxygen and 
sodium vapor; as soon as this dissociation takes place the 
carbon takes up the oxygen, forming carbonic oxide, and thus, 
by preventing the recombination of the sodium and oxygen, 
leaves free sodium vapors. 

Dr. Kosman, speaking in " Stahl und Eisen," January, 1889, 
on Castner's process, gives the following explanation of the re- 
actions taking place : — 

Ten kilos of caustic soda and 5 kilos of carbide (containing 
1.5 kilos of carbon) give the following reaction: 


4NaOH + FeCj = Na^COs + Fe + 4H + CO + 2Na, 

and half the sodium in the mixture is obtained. 

Ten kilos of caustic soda and 10 kilos of carbide (containing 
3 kilos of carbon) give this reaction — 

zNaOH + FeQ = NaCO + Fe + 2H + CO + Na, 

and half the sodium in the mixture is again obtained. 

If 20 kilos of caustic soda and 15 kilos of carbide are mixed, 
both the above reactions take place ; but if the ignition is con- 
tinued, the sodium carboxyd (NaCO) reacts on the sodium 
carbonate according to the reaction — 

Na^COs + NaCO = sNa + 2CO2, 

and the entire reaction may be represented by 

sNaOH + FeC^ = sNa + Fe + 3 H + CO -h CO,, 

and all the sodium in the mixture is obtained. 

This is the reaction first proposed by Mr. Castner (see p. 
206), and the proportions indicated by it gave him the largest 
return of sodium. Mr. MacTear, however, states that the re- 
action which takes place is conditioned largely by the tempera- 
ture, and that at 1000° C. it is probably to be represented by 

6NaOH -I- FeCj = 2Na2C03 + 6H -F Fe + zNa, 

which is essentially the same as that given by Sir Henry Ros- 
coe in his discourse, viz : — 

3NaOH + C = Na.CO, + 3H. + Na. 

This reaction would require 18^ lbs. of carbide to 50 lbs. of 
caustic soda, and since the sodium carbonate is easily converted 
back into caustic by treatment with lime, the production of so 
much carbonate is offset by the ease with which the reaction 
takes place, and the added advantage that the gas evolved with 
the sodium is solely hydrogen, thus allowing the reduction to 


proceed in an atmosphere of that gas, and reducing the pro- 
duction of the usual deleterious sodium carbides to a minimum. 
A further discussion of this subject will come up in consider- 
ing Netto's process. 

Netto's Process (1887). 
Dr. Curt Netto, of Dresden, took out patents in several 
Eurupean countries,* which were transferred to and operated 
by the AlHance Aluminium Company, of London (see p. 31). 
The process is continuous, and is based on the partial reduc- 
tion of caustic soda by carbon. Dr. Netto observes that car- 
bon will reduce caustic soda at first at a red heat, but a white 
heat is necessary to finish the reduction, the explanation being 
that the reaction is at first — 

4NaOH + C = Na^COa-l- 2H, -t- CO + Na„ 

and that the carbonate is only reduced at a white heat. To 
avoid any high temperature, the first reaction only is made use 
of, the carbonate being removed and fresh caustic supplied 
continuously, and without interrupting the operation or admit- 
ting air into the retort in which the reduction takes place. 

A vertical cast-iron retort, protected by fire-clay coating, is 
surrounded by flues. The flame after heating the retort passes 
under an iron pot in which the caustic soda is kept melted, and 
situated just above the top of the retort. This pot has an out- 
let tube controlled by a stop-cock, by which the caustic may 
be discharged into a funnel with syphon-shaped stem fastened 
into the top of the retort. There is also a syphon-shaped out- 
let at the bottom of the retort, through which the molten so- 
dium carbonate and bits of carbon pass. A hole with tight lid 
in the upper cover is provided for charging charcoal. A tube 
passes out just beneath the upper cover, connecting with a 
large condenser of the shape used by Deville (see Fig. 20). 
In operating, the retort is heated to bright redness, filled one- 
third with best wood charcoal, and then . molten caustic soda 

♦German patent (D. R. P.) 45105; English patent, October 26, 1887, No. 14602. 



tapped from the melting-pot into the funnel, the feed being so 
regulated that the funnel is kept full and the retort closed. 
The lower opening is kept closed until enough sodium carbon- 
ate has accumulated to lock the syphon passage air tight. 
When after several hours' working the charcoal is almost all 
used up, the supply of caustic soda is shut off for a time and 

Fig. 20. 

the retort recharged through the opening in the upper lid, 
when the operation goes on as before. The sodium carbonate 
produced is easily purified from carbon by solution. Since 
sodium vapor at a high temperature is very corrosive, all rivets 
and screw joints must be avoided in making the retort. On 
this account, the outlet tubes should be cast in one piece with 
the retort. 


Both Netto's process and Castner's were in use for a con- 
siderable time ( 1 888-1 891), and produced sodium in large 
quantities and cheaply. With the advent of the cheaper elec- 
trolytic methods of producing aluminium, the sodium methods 
were put to a sharp struggle for existence. The Alliance Alu- 
minium Company, working the Netto process, soon gave up 
the fight. The Aluminium Company, Limited, working Cast- 
ner's process, were enabled to continue making aluminium for 
some time longer by an invention of Mr. Castner's to produce 
sodium electrolytically. This process is still being used, but 
the sodium produced is used for various other purposes than 
making aluminium, the extraordinary cheapness of the electro- 
lytic processes rendering competition impossible. 

Reduction of Sodium Compounds by Electricity. 

The decomposition of fused sodium chloride by the electric 
current seems to promise the economic production of sodium, 
for not only is this metal formed, but chlorine is obtained as a 
by-product, its value reducing very much the cost of the 

P. JablochofF has devised the following apparatus for decom- 
posing sodium or potassium chlorides.* (Fig 21.) 

The arrangement is easily understood. The salt to be de- 
composed is fed in by the funnel into the kettle heated by a fire 
beneath. The positive pole evolves chlorine gas, and the 
negative pole evolves vapor of the metal, for, as the salt is 
melted, the heat is sufficient to vaporize the metal liberated. 
The gas escapes through one tube and the metallic vapor by 
the other. The vapor is led into a condenser and solidified. 

Prof. A. J. Rogers, of Milwaukee, Wis., has made a number 
of attempts to reduce sodium compounds electrolytically, using 
as a cathode a bath of molten lead and producing an alloy of 
lead and sodium which he makes use of for the reduction of 
aluminium compounds. Although these attempts are hardly 

* Mierzinski. 



past the experimental stage, yet the record of the results ob- 
tained may very probably be interesting and valuable to other 
investigators in this line. 

Prof. Rogers reasons that from the known heat of combina- 

FlG. 21. 

unvrm |>T«" 

tion of sodium and chlorine (4247 calories per kilo of sodium) 
there is enough potential energy in a pound of coal to separate 
nearly two pounds of sodium, if any mode of applying the com- 
bustion of the coal to this end without loss could be devised. 
If, however, this energy is converted into mechanical work, this 
again into electrical energy, and this latter used to decompose 
sodium chloride, we can easily compute the amount of coal to 
be used in a steam boiler to produce a given amount of sodium 
by electrolysis. Now, if the electric current could be applied 
without loss in decomposing sodium chloride, i electric horse- 
power (746 Watts) would produce about 8 lbs. of sodium in 
24 hours. But as in practice one mechanical horse-power ap- 
plied to a dynamo yields only 80 or 90 per cent, of an electric 
horse-power, and as about 4 lbs. of coal are used per indicated 


horse-power per hour, from 105 to 120 lbs. of coal would be 
required per day to produce this result, or about 15 lbs. per lb. 
of sodium. Since, however, there is a transfer resistance in the 
passage of the electric current through the molten electrolyte, 
more than this will be required, in proportion to the amount of 
current thus absorbed. 

The temperature of fusion of sodium chloride is given by 
Carnelly as Tj6° C, but Prof. Rogers remarks that the fusing 
point may be lowered considerably by the presence of other 
salts ; for instance, it melts about 200° lower if a small amount 
of calcium chloride or potassium chloride is present. We will 
quote the results of some experiments as given by Prof. 

"The following results were obtained among many others by 
using a Grove battery, a Battersea crucible to hold the sodium 
chloride, a carbon anode and an iron cathode terminating in a 
tube of lime placed in the melted salt. As soon as metallic 
sodium escaped and burnt at the surface of the liquid the cur- 
rent was stopped. A little sodium was oxidized, but a consid- 
erable amount was found in the tube in the metallic state. In 
six experiments the amount of sodium obtained was from 50 to 
85 per cent, of the theoretical amount, averaging 65 per cent. 
It thus seemed that, with suitable apparatus, from 5 to 6 lbs. of 
sodium could be obtained in 24 hours per electric horse-power. 
Thus, if there were no practical difficulties in the construction 
of the crucibles and other apparatus involved, nor in working 
continuously on a large scale, the metal could be obtained at 
small cost. Various forms of crucibles were used and attempts 
made to distil the metal when formed at the negative electrode 
(sodium volatilizing at about 900° C), but the sodium vapor 
carries with it a large amount of sodium chloride as vapor, and 
the distillation is attended with difficulty. 

" During the last three years I have experimented on the 
reduction of sodium chloride, using molten negative electrodes 
and especially lead. Lead, tin, zinc, cadmium and antimony 
* Proceedings of the Wisconsin Natural History Society, April, 1889. 


all readily .alloy, with sodium, a large part of which can be re- 
covered from the alloys by distillation in an iron crucible. They 
can be heated to a higher temperature than pure sodium in 
acid crucibles without the sodium attacking the crucible. In 
the following experiments a dynamo machine was used to sup- 
ply the current. 

"Experiment i. A current averaging 72 amperes and 33 
volts was passed through molten sodium chloride contained in 
two crucibles arranged in series, for two hours. Each con- 
tained 30 lbs. of salt; in the first was put 104 grammes of tin, 
in the second 470 grammes of lead, each serving as cathode, 
and connection being made through the bottom of the crucible. 
A carbon anode passed through the cover and extended to 
within three inches of the molten cathode. The crucible con- 
taining the tin was nearer the fire and consequently hotter, and 
had an average potential across the electrodes of 12 volts, 
while that containing the lead cathode was 21 volts. When at 
the end of two hours the carbons were removed and the cru- 
cibles cooled and broken open, the lead alloy was found to 
contain 96 grammes of sodium, or 17 per cent. There was 
about 90 grammes of sodium found in the tin alloy, or between 
45 and 50 per cent. Both these alloys rapidly oxidized in the 
air, and when thrown into water the action was very energetic, 
in the case of the tin alloy the liberated hydrogen being ignited, 
and after the reaction the metals were found at the bottom of 
the vessel in a finely divided state. Both these alloys reduce 
cryolite or aluminium chloride." 

In Prof. Rogers' further experiments cryoHte was added to 
the bath, so that sodium was produced and aluminium formed 
in one operation. (See under "Electrolytic Processes," 
Chap. XI.) 

L. Grabau, of the " Aluminium Werke zu Trotha," has 
made the following observations on the electrolysis of molten 
sodium chloride. When the melted, salt is subjected to the 
electric current, the resistance of the bath quickly rises and it 
becomes almost inipossible to force the current through, This 


is probably due to the formation of a non-conducting sub- 
chloride. The formation of this salt can be prevented by the 
addition of some potassium chloride and the chloride of an 
alkaHne-earth metal to the bath. In his patents,* Grabau 
recommends the use of a mixture of sodium, potassium and 
strontium chlorides, in the proportion of 88 sodium chloride, 
112 potassium chloride, and 159 strontium chloride. With this 
mixture, Grabau claims to obtain almost pure sodium, contain- 
ing only about 3 per cent, of potassium and no trace of 
strontium, while at the same time 95 per cent, of the metal 
which the current should theoretically give is obtained. For 
all the purposes to which sodium is now applied the presence of 
the potassium is not in the least prejudicial. Borchers sub- 
stantiates the fact that from a mixture of potassium and sodium 
chlorides, which is very fluid, only sodium is liberated by the 
current if it is not too strong and if the sodium chloride de- 
composed is at once replaced. For carrying on his process, 
Grabau uses the apparatus shown in Fig. 22. The positive 
poles are carbon rods (C), the negative an iron pipe E, slit in 
its lower part and surrounded by a bell-shaped porcelain cover 
B. As the molten sodium is lighter than the bath material, it 
rises into the pipe £ and passes away through the branch a into 
a condenser where it collects under petroleum. The screw H 
can be turned downwards to clean out the tube E in case it 
stops up. It is intended that the vessel should be heated by 
hot gases passing through the flues G G, and its temperature so 
regulated that the sodium passes off in the liquid state. The 
chlorine generated passes off by the tube d. In order to pre- 
vent the corrosion of the cover B, Grabau has even con- 
structed an apparatus in which it contained an iron lining and 
was water-cooled. 

The method now used by Castner .consists in electrolyzing 
molten caustic soda at the lowest possible temperature. Pure 
caustic soda melts below redness, and if kept at a temperature 

♦German Patents (D. R. P.) Nos. 51898 and 56230 (1890). 



not more than 20° above its melting point (310° C), it can be 
easily and continuously electrolyzed. Arrangements must 
therefore be provided for keeping the temperature of the bath 
very constant, and for removing the sodium quickly from the 
bath in order to prevent loss by corrosion. The apparatus 
consists of an iron crucible, embedded in brickwork, in which 

Fig. 22. 

the caustic soda is melted by the heat of a circular gas burner 
surrounding the pot (Fig. 23). The negative electrode passes 
up through the bottom of the pot, the space between it and the 
sides of the hole being allowed to fill up with solid caustic. A 
tubular sleeve of iron descends from the cover and envelopes 
the top of this electrode. Outside of this are the positive elec- 
trodes. An opening is provided in the cover for the escape of 



gas and introduction of a thermometer. In operation, the heat 
caused by the passage of the current is sufificient to keep the 
bath melted. The free sodium floats on the surface of the bath 

inside the iron sleeve surrounding the negative electrode, and 
is skimmed off with a small ladle having fine perforations. The 
operation is continuous, and afifords sodium at a cost probably 
not greater than 15 cents per pound. 



The branch of chemical science called thermal chemistry may 
be said to be yet in its infancy. Although an immense mass 
of thermal data has been accumulated, yet the era of great 
generalizations in this subject has not yet been reached ; and 
although we know with a fair degree of accuracy the heat of 
combination of thousands of chemical compounds, including 
nearly all the common ones, yet the proper way to use these 
data in predicting the possibility of any proposed reaction re- 
mains almost unknown. The principal barriers in the way are 
two: 1st, the unknown quantities entering into almost every 
chemical reaetion thermally considered, i. e., the heat of com- 
bination of elementary atoms to form molecules of the ele- 
ments ; 2d, the uncertainty as regards the critical temperature 
at which a given exchange of atoms and consequent reaction 
will take place. We will explain what is meant by these state- 

To illustrate, let us consider the case of hydrogen uniting 
with oxygen to form water according to the formula — 

2(H— H) -I- (0= 0)= 2H,0 

where (H — H) and (O ^ O) represent respectively molecules 
of hydrogen and oxygen. Now, as i kilo of hydrogen unites 
with 8 of oxygen to form 9 of water, setting free 34,462 units 
of heat (calories), if we take the atomic weights in the above 
reaction as representing kilos, we shall have the thermal value 
of the reaction 4 X 34,462 = -|- 137,848 calories. But this' 
quantity is evidently the algebraic sum of the heat evolved in 



the union of 4 kilos of hydrogen atoms with 32 kilos of oxygen 
atoms, and the heat absorbed in decomposing 4 kilos of hydro- 
gen gas into atoms! and 32 kilos of oxygen gas into atoms. 
These two latter quantities are unknown, though a few chem- 
ists have concluded from studies on this question that they are 
probably very large. It has been calculated that the reaction 

H -|- H = (H — H) sets free 240,000 calories, 
and 4-0^ (O = O) sets free 147,200 calories ; 

If these numbers are approximately true, then 

4H + 20^ 2H2O would set free about 773,000 calories. 

If these quantities are really anything like so large, and if 
they are at sometime determined with precision, thermo-chem- 
ical principles and conclusions will be greatly modified. Mean- 
while, predictions based on the data we have, lose all possibility 
of certainty, and so we need to keep in mind in our further 
discussion that our deductions at the best are only probabil- 
ities. Further, suppose that we mix i kilo of hydrogen gas 
and 8 kilos of oxygen gas, put them in a tight vessel and keep 
them at the ordinary temperature. No reaction will take place 
in any length of time, even though 34,462 calories would be 
set free thereby. The explanation of this is that the atoms of 
hydrogen and oxygen are so firmly bound to each other in the 
molecules, that the dissimilar atoms have not strength of afiSnity 
sufiScient to break away in order to combine. It is well known 
that a spark only is necessary to cause an explosive combina- 
tion of the gases under the above conditions, the temperature 
of the spark expanding the gases coming in contact with it, 
causing the atoms to swing with more freedom in the molecules ; 
and as soon as two atoms of hydrogen come within the .sphere 
of attraction of an atom of oxygen and form a molecule of 
water, the heat liberated is immediately communicated to the 
adjacent atoms, and almost instantaneously the entire gases 
have combined. The same principle undoubtedly holds true 


in cases of reduction. Carbon may be mixed with litharge 
and the mixture left in the cold forever without reacting, but at 
a certain temperature the carbon will abstract the oxygen. The 
temperatures at which reactions of this nature will take place 
are often determined experimentally. 

There are other points which are somewhat indeterminate in 
these discussions, such as the influence of the relative masses 
of the reacting bodies, their physical states, i. e., solid, liquid 
or gaseous, also the influence of the physical conditions favor- 
ing the formation of a certain compound ; but the nature of the 
subject and the meagreness of data in the particular pheno- 
menon of reduction, render it inexpedient if not impracticable 
to take these points into consideration. 

Starting with the above remarks in view, we will consider the 
heat generated by the combination of aluminium with certain 
other elements, as has been determined experimentally, and 
study, from a comparison with the corresponding thermal data 
for other elements, what possibilities are shown for reducing 
these aluminium compounds. 

The heat generated by the combination of aluminium with 
the different elements is given as follows : the first column giv- 
ing the heat developed by 54 kilos of aluminium (representing 
AI2), and the second the heat per atomic weight of the other 
element, e. g., per 16 kilos of oxygen. 

Element. Compound. Calories. Calorie.s. Authority. 

Oxygen * Al^Oj 391,600 130,500 Berthelot. 

392,600 130,900 Bailie & F^ry. 

Fluorine 2AIF3 552,000 92,000 Berthelot. 

Chlorine 2AICI3 321,960 53,66o Thomsen. 

Bromine 2AlBr3 239,440 39,900 " 

Iodine 2AU3 140,780 23,460 " 

Sulphur AI2S3 124,400 41,467 Sabatier. 

* Berthelot's number represented the formation of the hydrated oxide, AIJO3.3H2O, 
and, for want of knowing the heat of hydration, has been generally used as the heat 
of formation of AljOj. Recently, J. B. Bailie and C. F4ry (Ann. de Chim. et de 
Phys., June, 1889, p. 250) have, by oxidizing aluminium amalgam, obtained the above 
figure for the heat of formation of AI2O3, and determined that the beat of hydration 
is 3000 calories, which would make the heat of formation of the hydrated oxide 


Let US consider the theoretical aspect of the reduction of 
alumina. The heat given out by other elements or com- 
pounds which unite energetically with oxygen is as follows, the 
quantity given being that developed by combination with 16 
kilos (representing one atomic weight) of oxygen. 

Element. Compound. Calories. 

AlTiminlum AljOj 130,500 

Sodium Na.p 99, 760 

Potassium Kfi 100,000 ( ?) 

Barium BaO 124,240 

Strontium SrO 1 28,440 

Calcium CaO 130,930 

Magnesium MgO 145,860 

Manganese MnO 95,ooo ( ?) 

Silicon SiOj 1 10,000 

Zinc ZnO 85,430 

Iron FejOj 63,700 

Lead PbO 50,300 

Copper CuO 37>i6o 

" CU2O 40,810 

Sulphur SO2 35.540 

Hydrogen H^O 68,360 

Carbon CO 29,000 

" COj 48,480 

Carbonic anhydride CO^ 67,960 

Potassium cyanide KCyO 72,000 

On inspecting this list we find magnesium to be the only metal 
surpassing aluminium, while calcium is about the same. This 
would indicate that the reaction 

AlA + 3Mg = Al, + 3MgO 

would, if it were possible to bring the alumina and magnesium 
in the proper conditions for reacting, develop about 

(145,860 — 130,500) X 3 = 46,080 calories, 

and points to the possibility of reducing alumina by nascent, 
molten, or vaporized magnesium, under certain conditions. In 
fact, since this paragraph was first written, in 1890, Dr. Clemens 
Winkler has succeeded in performing the reduction in part.* 

* Berliner Berichte, 1890, p. 772. 


He found that when alumina was heated with magnesium pow- 
der in a current of hydrogen, a nearly black powder resulted, 
containing a lower oxide of aluminium, AlO, the reaction 

2AI2O, + Mg = MgO.AljO, + 2AIO ; 

the magnesia formed uniting with alumina to form an alum- 

We notice further the fact that sodium or potassium could 
not reduce alumina without heat being absorbed in large 
quantity, and it is interesting to remember that some of the 
first attempts at isolating aluminium by using potassium were 
made on alumina, and were unsuccessful, so that it is practi- 
cally acknowledged that while these metals easily reduce other 
aluminium compounds (according to reactions which are ther- 
mally possible, as we shall see later on) yet they cannot reduce 
alumina, under any conditions so far tried. 

When we consider the case of reduction by the ordinary re- 
ducing agents, hydrogen, carbon, or potassium cyanide, we are 
confronted in every case with large negative quantities of heat, 
i. e., deficits of heat. So large do these quantities appear that 
it is very small wonder that the impossibility of these reduc- 
tions occurring under any conditions has been strongly affirmed, 
For instance 


would require 

(130,500 — 68,360) X 3 = 186,420 calories. 

AI2O3 + 3C =Alj -I-3CO 304,500 calories. 

AI2O3 + i>^C = AI2 -h ii^COa 246,060 calories. 

Al,03 -I- 3KCy = Al, + 3KCyO 175,500 calories. 

AlA +3CO =h\ + 7,C0, 187,620 calories. 

The above quantities represent the differences between the 
heats of formation of alumina and the product of the reduc- 
tion, assuming the reaction to take place at ordinary temper- 
atures. At high temperatures these heats of formation are in 


general less than they would be at low heats, and if the heat of 
formation of alumina decreases faster than .the heat of combi- 
nation of the reducing agent with oxygen, the dififerences noted 
above would be less and the reaction easier to produce. This 
must be what actually takes place. For instance, at the heat 
of an electric furnace, say 2500° to 3000°C., molten alumina 
is reduced by carbon present in large quantity, that is, the dif- 
ference beiween the heats of formation of AI2O3 and 3 CO has 
become so small that the reaction takes place. Similarly, 
Warren has recently reduced alumina by heating it in a lime 
tube by an oxy-hydrogen blow-pipe, passing hydrogen through 
the tube. We here have a reaction which is negative by 186,- 
000 calories at ordinary temperatures, taking place at a tem- 
perature probably not over 2000°C. 

The conditions which render the carrying out of these diffi- 
cult reductions possible are therefore : 

1 . A high temperature, which weakens the heat of formation 
of alumina. 

2. An excess of the reducing agent, which is thus assisted 
simply by the overpowering influence of its mass. 

3. Fluidity of the substance to be reduced. 

4. Intimate contact with the reducing agent. 

5. Favorable conditions for the formation and removal of the 
gaseous products of the reduction. 

Concerning these, I would remark that the fluidity of the sub- 
stance to be reduced may be attained by other means than a 
high temperature. The alumina may, for instance, be liquefied 
by solution in a bath of molten cryolite. If we dissolve it in 
molten caustic soda we get a combination of alumina and soda 
which will remain fluid if too much alumina is not added. 
Such fluidity also allows much more intimate contact with the 
reducing agent if it is a solid. . If the reducing agent is a gas, 
most intimate contact is obtalined by finely pulverizing the alu- 
mina. If the reducing agent is a solid and the product of the 
reduction would be a gas, its formation would be accel- 
erated by reducing the pressure in the apparatus. In other 



words, a partial vacuum would undoubtedly aid in bringing 
about the reduction, and would help to keep the aluminium in 
the metallic state by removing quickly the product of reduction 
and so preventing re-oxidation. I have no hesitation in affirm- 
ing that since we now know that carbon does reduce molten 
alumina below 3000°, and hydrogen reduces solid alumina at 
2000°, the application of correct principles by intelligent inves- 
tigators, with properly constructed apparatus, would undoubt- 
edly result in discovering methods of producing the same reac- 
tions at lower temperatures and on a larger scale. Whether 
these methods of producing aluminium could compete with the 
present electrolytic ones is another question, but it is certainly 
interesting to mark the possibility of processes which some 
future investigator may bring to commercial success. 

As the basis of our discussion of the reduction of aluminium 
chloride, bromide, or iodide, we give a table of heat developed 
by the combination of some of the elements with one atomic 
weight (in kilos) of each of these haloids. 































f 5o,6oot 
1 48,600* 


] 40,640t 
I 37.500* 
























Potassium . . 


Lithium .... 
Barium .... 
Strontium . . 
Calcium .... 
Manganese . 



Mercury . . . • 



Copper . . . . 
Hydrogen. . 


















f 30,coot 

■j 26,600 

(. 24,500 



— 6,000 

*Thomsen. t Andrews. J Jahresbericht der Chemie, 1878, p. 102, 

On inspecting this table we notice that, in general, all the 


metals down to zinc develop more heat in forming chlorides, and 
very probably also in forming bromides and iodides, A re- 
action, then, between aluminium chloride and any of these 
metals, forming aluminium and a chloride of the metal, would 
be exothermic, which means, generally speaking, that if alu- 
minium chloride and any one of these metals were heated to- 
gether to the critical point at which the reaction could begin, 
the reaction would then proceed of itself, being continued by the 
heat given out by the first portions which reacted. Zinc seems 
to lie on the border line, and the evidence as to whether zinc 
will practically reduce these aluminium compounds is still con- 
tradictory, as may be seen by examining the paragraphs under 
" Reduction by Zinc." (Chap. XII.) 

Of the first six metals mentioned in the table after aluminium, 
only potassium and sodium are practically available. The re- 

AlCl3 + 3K = AH-3KCl develops 155,820 cal. 
AlCls-F 3Na = Al-f- 3NaCl develops 132,090 cal. 

and the result of this strong disengagement of heat is seen 
when, on warming these ingredients together, the reaction once 
commenced at a single spot all external heat can be cut off, and 
the resulting fusion will become almost white hot with the heat 
developed. In fact, the heat developed in the second reaction 
would theoretically be sufficient to heat the aluminium and 
sodium chloride produced to a temperature between 3000° and 

Magnesia should act in a similar manner, though not so vio- 
lently, since 

2AlCl3 + 3Mg=Al2 + 3MgCl2 develops 131,000 cal. 

And manganese possibly also, since 

2AICI3 + 3Mn = Ala -f- 3MnCl develops 14,040 cal. 

The reduction of aluminium chloride, bromide, or iodide by 
hydrogen is thermally strongly negative, which would indicate 
a very small possibility of the conditions ever being arranged so 


as to render the reaction possible. For instance, taking the 
most probable case, 

AICI3 + 3H = Al + 3HCI requires 94,980 calories. 

The only probable substitutes for sodium in reducing alumin- 
ium chloride are thus seen to be magnesium (whose cost will 
probably be always greater than that of aluminium), manganese 
(which may sometime be used in the form of ferro-manganese 
for producing ferro-aluminium), and zinc (whose successful 
application to this purpose would be a most promising advance 
in the metallurgy of aluminium). 

In 1886 Moissan isolated fluorine, and in 1889 he and Berthe- 
lot measured the heat of formation of hydrofluoric acid. With 
this datum, it became possible to calculate a table of the heat 
of formation of the fluorides.* These can now be given as 
follows, stating the heat evolved by combination with one 
atomic weight (19 kilos) of fluorine. 

Element. Compound, Heat evolved. 

Potassium KF 1 10,000 

Sodium NaF 109,700 

Barium BaF^ 112,000 

Strontium SrFj 112,000 

Calcium CaF, 108,000 

Magnesium MgFj 105,000 

Aluminium AIF3 92,000 

Manganese MnFj 72,000 

Zinc ZnFg 69,000 

Cadmium CdF^ 61,000 

Iron FeF^ 63,000 

Cobalt C0F2 60,000 

Nickel NiFj 59,500 

Copper CuFj 44,000 

Silver AgF 25,500 

Hydrogen HF (gas) 37,6oo 

" " (in solution) 49,400 

It follows from an inspection of these quantities that the 
fluorides are stronger compounds than either the oxides, chlor- 
ides, bromides or iodides. It is also seen that aluminium 

* See a paper on this subject, by the author, in the Journal of the Franklin Insti- 
tute, June, 1891. 


stands in the same relative position as in the chlorides, except 
that it is here stronger than manganese or zinc. It follows that 
only sodium, potassium or magnesium of the available metals 
wilU reduce aluminium fluoride. For such reductions the latter 
must be liquefied by mixture with a fusible salt not acted on by 
the reducing agent. Thus, we* would have 

2AIF3 + 6Na= 6NaF+ 2AI, evolving 106,200 calories. 
2A1F3 + 3Mg = 3MgF2 + 2AI, evolving 78,000 calories. 
The reaction with zinc would absorb heat as follows : 
2AIF3 + 2Zn= 2ZnF2 + 2 Al, absorbing 138,000 calories. 

While the first two reactions take place easily, at a red heat, 
the latter is nearly impracticable. The writer brought zinc into 
contact with highly heated cryolite (AlFj.sNaF), and obtained 
0.6 per cent, of aluminium in the zinc* The temperature was 
a bright red, and this small amount of reduction shows that an 
excess of zinc at a high temperature will to a small extent over- 
come the affinity of aluminium for fluorine. Manganese in the 
form of ferro-manganese, might take up even a larger percent- 
age of aluminium than the above, yet no substantial amount of 
reduction can be looked for from these elements. 

In order to discuss the thermal relations of aluminium sul- 
phide, we will make use of the following data, the heat devel- 
oped being per atomic weight (32 kilos) of sulphur combining: 

Element. Compound. Calories. 

Alumlnitiin AI283 41,467 

Potassium KjS 103,700 

Sodium Na^S 88,200 

Calcium CaS 92,000 

Strontium SrS 99,200 

Magnesium MgS 79,600 

Manganese MnS 46,400 

Zinc ZnS 41.326 

Iron FeS 23,576 

Copper Cu^S 20,270 

Lead • PbS 20,430 

Hydrogen HjS 4>740 

Carbon CSj —14,500 

* See details in Chapter XII. 


These figures point to the easy reduction of aluminium sul- 
phide by potassium, sOdium, or magnesium, and its possible 
reduction by manganese and zinc. The other metals would 
require exceptional conditions, perhaps, of temperature, for 
their action. It is interesting to note that two observers have 
determined from the deposition -of the metals from solution by 
hydrogen sulphide, that the order of the afifiinity of the metals 
for sulphur is first the alkaline metals, then the others in the 
following order: copper, lead, zinc, iron, manganese, and then 
aluminium and magnesium — with the remark that the affiinities 
of the latter two for sulphur appear quite insignificant. The 
explanation of this difference is that the metals last mentioned 
form sulphides which are not stable in water, but are immedi- 
ately decomposed according to the reaction 

3 AI2S3 + 3 H2O = AI2O3 + 3H2S, evolving 77,840 calories, 

while the alumina produced is immediately soluble in the weak 
acid solution and at once disappears. It follows that these ele- 
ments are not precipitated as easily from solution as those 
whose sulphides are not thus decomposed by water. 

The reduction of aluminium sulphide by copper, tin and iron 
is in a measure possible when the operation is conducted at a 
high temperature and there is an excess of reducing metal 
present. Under these conditions, a part of the aluminium sul- 
phide is reduced and an alloy formed, but such reactions can- 
not approach in completeness those reactions which are 
strongly endothermic. The reduction by hydrogen is seen to 
be highly improbable, and by carbon still more so. 

The question of reducing aluminium compounds by the use of 
electricity is a very different question. Looked at as a reduc- 
ing or decomposing agent, the electric current is almighty. A 
current with a tension of i volt will decompose any fluoride, 
chloride, bromide or iodide whose heat of combination with one 
atom weight of fluorine is less than 23,000 calories ; at a tension 
of two volts, any less than 46,000 calories ; and therefore at a 
tension of five volts would decompose the strongest of any of 


these salts. Such a current only means half a dozen ordinary 
copper sulphate batteries connected in tension. For oxides 
and sulphides, the heat of combination with one-half an atomic 
weight must be used, which quantity is equal in combining 
power to one whole atomic weight of the above elements, and 
then the same rule applies. We can thus calculate the voltage 
required to decompose the following anhydrous molten com- 
pounds — 

» Element. Fluoride. Chloride. Oxide. Sulphide. 

Potassium 4.8 4.6 2.2 2.3 

Sodium 4.7 4.3 2.2 1 .9 

Barium 4.9 4.3 2.7 2.2 

Strontium 4.9 4.0 2.75 2.2 

Calcium 4.7 3.7 2.85 2.0 

Magnesium 4.6 3.3 3.2 1.7 

Aluminiuin 4.0 2.3 2.8 0.9 

Manganese 3.1 2.4 2.1 l.o 

Zinc 3.0 2.3 1 .9 0.9 

Iron 2.7 1.8 1.5 0.5 

Nickel 2.5 1.6 1.3 0.4 

Copper 1.9 1.4 0.9 0.4 

Silver i.i 1.3 o.l o.i 

Hydrogen 1.6 i.o 1.5 0.1 

A critical comparison of these figures leads to some very in- 
teresting conclusions. We notice, in the first place, that in the 
fluorides there is a very sharp drop after aluminium ; in these 
salts aluminium has closer resemblance to the alkaline earth 
fluorides, while manganese and zinc resemble the ordinary 
metals. In the chlorides, there is a sharp drop before alumin- 
ium, leaving it in stronger resemblance to the zinc, manganese 
and lower metallic salts. In the oxides, aluminium is again 
classed with the alkaline earth metals, and sharply separated 
from the lower metallic oxides. Attention is particularly called 
to the irregularity of the alkaline metals, whose oxides are as 
easily decomposed as those of the metals below aluminium. 
This explains why we can reduce these alkaline metals more 
easily by carbon than we can alumina, and yet we can use the 
alkaline metal to react on a salt of aluminium and get metallic 


aluminium. The high voltage required to decompose mag- 
nesium oxide also explains the wonderful efficiency of mag- 
nesium as a reducing a:gent when acting on the oxides. In the 
list of sulphides, we again notice a sharp drop before alumin- 
ium, leaving it in close correspondance with manganese and 
zinc. The very low voltage required is also noticeable ; a 
single Daniell cell should be able to produce decomposition. 
Take a bath, for instance, in which alumina is being decom- 
posed using a tension of 5 volts, of which 2.8 is absorbed in 
decomposition and 2.2 in overcoming other resistances. If 
aluminium sulphide were the material decomposed, assuming 
other resistances the same as before, the total resistance of the 
bath would be overcome by 3.1 volts; or, in other words, it 
would require only 62 per cent, of the power to produce the 
same weight of aluminium. The sulphide thus presents great 
advantage in respect to ease of decomposition. 

Next to the sulphide, the chloride is the easiest to decom- 
pose, but both these salts are difficult to produce, and herein 
lies their chief disadvantage. 

It can be furthermore observed, that if we mix these various 
salts into one bath, and use an electric current carefully regu- 
lated as to tension, we can act on pne ingredient without dis- 
turbing to any extent the others present. We could, for 
instance, mix aluminium chloride with any fluoride down to 
that of zinc, and by a tension of 2.5 volts or a little over sepa- 
rate out only alumininum. We could mix alumina with any 
fluoride or chloride above aluminium in the list, and get only 
aluminium from the bath. The fact is, however, that alumina 
is not dissolved by the chlorides, but only by the fluorides. 
These latter are therefore available, and are the principal salts 
used at present in the aluminium industry. Thus, in Hall's 
process we have present in the bath 

Alumina requiring 2.8 volts for decomposition. 

Aluminiunl fluoride " 4.0 " " 

Sodium '■ " 4.7 . " , •' 

When the current is passed through such a mixture, and its 


tension does not rise abnormally high,' the bath being kept sat- 
urated with all the aluimina it can dissolve, the only ingredient 
decomposed is the alumina. 

In Minet's process there are present : 

Alumina, requiring 2.8 volts. 

Aluminium fluoride, " 4.0 " 

Sodium fluoride, " 4^7 " 

Sodium chloride, " 4.3 " 

The minimum electro-motive force required to decompose 
this bath at 900° is 2.4 volts; at 1100°, 2.17 volts. The elec- 
tro-motive force of carbon burning to carbonic oxide at those 
temperatures is, however, 0.65 volts, so that the voltage re- 
required to decompose liquid alumina at these temperatures is 
0.65 volts higher than the figure given. We therefore have 
these data for the decomposition of alumina : 

Voltage for 



Heat of Formation. 



392,600 (solid aluminium.) 



387,200 (liquid aluminium). 



424,600 (liquid aluminium). 



392,600 (liquid aluminium). 

The equation for the voltage required to decompose alumina 
producing liquid aluminium is therefore : 

V^ 2.78 + 0.00129 t — 0.00000128 1'', 

and the heat required to decompose the molecule of alumina is 

Q = 387,200+ 192 t — 0.1766 1^ 

This formula is extremely interesting from a thermo-chemi- 
cal point of view in enabling us to calculate theoretically the 
temperature at which hydrogen or carbon begins to reduce alu- 
mina. The heat of formation of gaseous water ,at zero is accu- 
rately known, and the experiments of Mallard and Le Chatelier 
on the specific heats of hydrogen, oxygen and water vapor 
enable us to calculate the heat of formation at any desired tern- 


perature. From these data we calculate the heat of formation 
of 3HjO to be represented by the formula: 

Q = 174,300+ 7.71 t — 0.0072 t\ 

Now to find the temperature at which hydrogen gas would be- 
gin to reduce alumina, we equate this formula with the one 
given for the heat of the formation of alumina, and have 

387,200 + 192 t — 0.1766 t''= 174,300+ 7.71 t— 0.0072 t'' 

from which 

t = i785°C. 

This will be the temperature at which the heat of formation 
of 3H2O will be exactly equal to the heat of formation of Al^Oj, 
and consequently immediately above this temperature reduc- 
tion will begin if the physical conditions are favorable for the 
reaction taking place. 

On a similar principle we can calculate the temperature at 
which carbon will begin to reduce aluminium, except that the 
operation is more difficult because of the variable specific heat 
of carbon. I have calculated from Weber's results that the heat 
in carbon from zero to high temperatures may be expressed as 

Q = o.5 t — 120. 

Taking this quantity, in connection with the specific heats of 
oxygen and carbonic oxide as determined by Mallard and Le 
Chatelier at high temperatures, we have for the heat of forma- 
tion of 3CO at any temperature t above 1000°, the expression 

Q = 83.35s + 8.6472 t - 0.0009 t^ 

For finding the temperature at which reduction begins we have 

387,200 + 192 t — 0.1716 t'' = 83,355 + 8.6472 t — 0.0009 ^^ 

from which 

t = 1980°. 


The temperatures above calculated appear much lower than 
has heretofore been deemed possible, but they are supported by 
experimental evidence ; viz., the reduction of alumina by 
hydrogen performed by Warren, at a temperature which it is 
hardly possible could have been higher than 2000° ; and the 
production of pig-iron containing over i per cent, of aluminium 
in a Pennsylvania blast furnace. The highest temperature at- 
tained in this furnace, not of very modern construction, could 
not have exceed 2000° C. (For details, see Chapter XII.) 

If in place of free hydrogen gas or free carbon we use as a 
reducing agent a combination of these two which gives out heat 
in its dissociation into carbon and hydrogen, we would have a 
more efficient reducing agent than if the constituents were used 
separately. Such a compound is Acetylene, CjHj, which ab- 
sorbs heat in its formation and therefore gives out heat in its 
decomposition. At ordinary temperatures, Berthelot has found 

C2 + H2 = C.2H.2 absorbs 5 1,500 calories. 

We do not know exactly how this number varies with the tem- 
perature, as we lack the specific heat of CJ-Ij, but we know that 
it does not change very much. We would therefore have the 
reduction equation as follows — 

AI2O3 + C.,H2 = 2AI + 2CO -I- H,0 

and the heat requirements as follows — 

Reducing AljOa ■• 387,200 -|- 192.61— 0.17161'' (l) 

Developed in forming 2CO 55.57° -f- S-76t — o.ooo6t'' (2) 

" " « HjO 58,000-1- 2.571— 0.00251^ C3> 

" " decomposing CijHj. 51,500 (4) 

If we put (i) = (2) + (3) + (4) and solve for t, we obtain 
the theoretical temperature at which Acetylene should begin to 
reduce alumina as 


This temperature, it will be observed, is not so low as that 


calculated for pure hydrogen. Six months ago, this calculation 
would have been of purely scientific interest, because pure 
acetylene was a difiScult substance to procure ; but a very prac- 
tical interest has been given it by the recent invention of Mr. 
Thomas L. Willson of Brooklyn*, whereby pure acetylene can 
be manufactured at a cost approximating 2 cents per pound. 
Since the gas should theoretically reduce double its weight of 
aluminium, if perfectly utilized for reduction, the cost of the re- 
ducing agent per pound of aluminium would be very small. It 
is to be hoped that the possibility of the above reaction may be 
tested with this powerful reducing agent. 

Before closing this subject of the thermal aspect of the reduc- 
tion of aluminium compounds, it may be interesting to notice 
some of the reactions which are of use in the aluminium indus- 
try. It is well known that while chlorine gas can be passed 
over ignited alumina without forming aluminium chloride, and 
while carbon can be in contact with alumina at a white heat 
without reducing it, yet the concurrent action of chlorine and 
carbon will change the alumina into its chloride, a compound 
with a lower heat of formation. Thus — 

AL^Os -f- 3C = AI2 4- 3CO requires 304,500 cal. 

But AI2O3 + 3C + 6C1 = 2AICI2 + 3CO requires a quantity of 
heat equal to the 304,500 cal. minus 321,960, the heat of for- 
mation of aluminium chloride, or in other words, 17,360 cal. is 
evolved, showing that the reaction is one of easy practicability. 
If it be inquired whether there is not some chloride which 
would act on alumina to convert it into chloride, we would re- 
mark that if we can find a chloride whose heat of formation is 
as much greater than the heat of formation of the correspond- 
ing oxide as the heat of formation of aluminium chloride is 
greater than that of alumina, then such a chloride might react. 

* See description in Engineering and Mining Journal, Dec. 15, 1894. 


To be more particular, to convert alumina into aluminium 
chloride, a deficit of 391,600 — 321,960 = 69,640 calories must 
be made up. If we know of an element which in uniting with 
3 atom weights (48 kilos) of oxygen gives out 69,640 calories 
more heat, or a still greater excess, than in uniting with 6 atom 
weights (213 kilos) of chlorine, then the chloride of that ele- 
ment might perform the reaction. Now — 

6Na + 3O = sNajO evolves 299,280 cal. 
and 6Na + 6C1 =6NaCl evolves 586,140 cal. 

leaving evidently a balance of 286,860 calories in the opposite 
direction to what we are looking for. And so for every metal 
except aluminium, I find the heat of formation of its chloride 
greater than an equivalent quantity of its oxide. The only 
element which I know to possess the opposite property is 
hydrogen, for — 

6H + 3O — 3H2O evolves 205,080 cal. 
and 6H + 6C1 = 6HC1 evolves 132,000 cal. 

and therefore the reaction — 

AI2O3 + 6HC1 = 2AICI3 + 3H2C 

would evolve according to our calculations (205, 080 — 132,000) — 
69,640=3,440 calories, and would be thermally considered a pos- 
sible reaction. Moreover, as a secondary effect, the water formed 
is immediately seized by the aluminium chloride, for the reaction 

2AICI3 + 3H2O = 2AICI3.3H2O evolves 153,690 cal. 

and thus increases the total heat developed in the decom- 
position of the alumina to 158, [30 calories. The result of 
this reaction is therefore the hydrated chloride, which is of no 
value for reduction by sodium, since when heated it decom- 
poses into alumina and hydrochloric acid again, that is, it will 
decompose before giving up its water, and the water if unde- 
composed, or the acid if it decomposes, simply unites with the 


sodium without affecting the alumina. The immense heat of 
hydration, 153,690 calories, is so much greater than any other 
known substance, that it is vain that we seek for any material 
which might abstract the water and leave anhydrous aluminium 

Analagous to the reaction by which aluminium chloride is 
formed from alumina is the reaction made use of for obtaining 
aluminium sulphide, yet with some thermal considerations of a 
different and highly interesting kind. If a mixture of alumina 
and carbon is ignited and, instead of chlorine, sulphur vapor is 
passed over it, no aluminium sulphide will be formed. An ex- 
planation of this fact is seen on discussing the proposed reac- 
tion thermally. 

AI2O3 + 3C -I- 3S = AI2S3 + 3CO requires 180,200 cal. 

It will be remembered that the similar reaction with chlorine 
evolved 17,360 calories; the quantity causing this difference is 
the heat of combination of aluminium sulphide, which is 321,960 
— 124,400 = 197,560 calories less than that of aluminium 
chloride, changing the excess of 17,360 calories into a deficit of 
180,200 calories. This large negative quantity shows a priori 
that the reaction could be made to occur only under excep- 
tional conditions, and its non-occurrence under all conditions 
so far tried gives evidence of the utility of the study of thermo- 
chemistry, at least as a guide to experiment. However, while 
carbon and sulphur cannot convert alumina into aluminium sul- 
phide, carbon bisulphide can, for a current of the latter led over 
ignited alumina converts it into aluminium sulphide. The re- 
action taking place is 

Al^Oa + 3CS2 = A1,S3 + 3COS. 

Now, since carbon and sulphur by themselves could not per- 
form the reaction, we should be very apt to reason that a com- 
pound of carbon and sulphur would be still less able to do so, 
since the heat absorbed in dissociating the carbon-sulphur 


compound would cause a still greater deficit of heat. But here 
-is precisely the explanation of the paradox. Carbon bisulphide 
is one of those compounds, not frequent, which has a negative 
heat of formation ( — 29,000 calories), i. e., heat is absorbed in 
large quantity in its formation, and therefore, per contra, heat 
is given out in the same quantity in its decomposition. The 
heat of formation of carbon oxysulphide being 37,030 calories, 
we can easily compute the thermal value of the reaction just 

Ifeat absorbed. 

Decomposition of alumina 391,600 cal. 

Heat developed. 

Decomposition of carbon bisulphide 87,000 

Formation of carbon oxysulphide 1 1 1,090 

" of aluminium sulphide 124,400 

322,490 cal. 

Deficit of heat 69,1 10 " 

It is thus seen that the reaction with carbon bisulphide is 
less than one-half as strongly negative as the reaction with car- 
bon and sulphur alone, and we therefore have an explanation 
of the fact that the reaction with carbon bisulphide is easily 
practicable, while the other is not. We do not know enough 
about the variation of the heats of formation of aluminium sul- 
phide and carbon oxysulphide to be able to make calculations 
as accurately as was possible with the question of reducing alu- 
mina, but if all the heats of formation remained constant except 
that of alumina, the 69,000 calories deficit would be made up 
by the decrease in the heat of formation of chat substance at 
975O C, which is somewhere about the temperature at which 
the reaction really takes place. 



The methods comprised under this heading may be con- 
veniently divided into two classes : — 

I. Methods based on the reduction of aluminium chloride 

or aluminium-sodium chloride. 
II. Methods based on the reduction of cryolite or aluminium 


The methods here included can be most logically presented 
by taking them in chronological order. 

Oerstedt's Experiments (1824). 

After Davy's unsuccessful attempts to isolate aluminium by 
the battery, in 1807, the next chemist to publish an account of 
attempts in this direction was Oerstedt, who published a paper 
in 1824 in a Swedish periodical.* Oerstedt's original paper is 
thus translated into Berzelius's " Jahresbericht :"f 

" Oerstedt mixes calcined and pure alumina, quite freshly 
prepared, with powdered charcoal, puts it in a porcelain retort, 
ignites and leads chlorine gas through. The coal then reduces 
the alumina, and there results aluminium chloride and carbonic 
oxide, and perhaps also some phosgene, COCI2 ; the alumin- 
ium chloride is caught in the condenser and the gases escape. 
The sublimate is white, crystalline, melts about the tempera- 
ture of boiling water, easily attracts moisture, and evolves heat 

* Oversigt over det K. Danske Videnskabemes Selkabs Forhandlingar og dets 
Medlemmers Arbeider. May, 1824, to May, 1825, p. 15. 
fBerz. Jahresb. der Chemie, 1827, vi. 1 1 8. 



when in contact with water. If it is mixed with a concentrated 
potassium amalgam and heated quickly, it is transformed ; there 
results potassium chloride, and the aluminium unites with the 
mercury. The new amalgam oxidizes in the air very quickly, 
and gives as residue when distilled in a vacuum a lump of metal 
resembling tin in color and lustre. In addition, Oerstedt found 
many remarkable properties of the metal and of the amalgam, 
but he holds them for a future communication after further in- 

Oerstedt did not publish any other paper, and the next ad- 
vance in the science is credited to Wohler, whom all agree in 
naming as the true discoverer of the metal. 

Wohler' s Experiments (1827). 

In the following article from Poggendorff 's Annalen,* Wohler 
reviews the article of Oerstedt's given above, and continues as 
follows : — 

" I have repeated this experiment of Oerstedt, but achieved 
no very satisfactory result. By heating potassium amalgam 
with aluminium chloride and distilling the product, there re- 
mained behind a gray, melted mass of metal, but which, by 
raising the heat to redness, went ofif as green vapor and distilled 
as pure potassium. I have therefore looked around for another 
method or way of conducting the operation, but, unpleasant as 
it is to say it, the reduction of the aluminium fails each time. 
Since, however, Herr Oerstedt remarks at the end of his paper 
that he did not regard his investigations in aluminium as yet 
ended, and already several years have passed since then, it 
looks as if I had taken up one of those researches begun aus- 
piciously by another (but not finished by him), because it 
promised new and splendid results. I must remark, however, 
that Herr Oerstedt has indirectly by his silence encouraged me 
to try to attain to further results myself. Before I give the art 
how one can quite easily reduce the metal, I will say a few 
words about aluminium chloride and its production (see p. 152). 

* Pogg. Ann., 1827, ii. 147. 


" I based the method of reducing aluminium on the reaction 
of aluminium chloride on potassium, and on the property of the 
metal not to oxidize in water. I warmed in a glass retort a 
small piece of the aluminium salt with some potassium, and the 
retort was shattered with a strong explosion. I tried then to do 
it in a small platinum crucible, in which it succeeded very well. 
The reaction is always so violent that the cover must be weighted 
down, or it will be blown off; and at the moment of reduction, 
although the crucible be only feebly heated from outside, it 
suddenly glows inside, and the platinum is almost torn by the 
sudden shocks. In order to avoid any mixture of platinum 
with the reduced aluminium, I next made the reduction in a 
porcelain crucible, and succeeded then in the following manner : 
Put in the bottom of the crucible a piece of potassium free from 
carbon and oil, and cover this with an equal volume of pieces 
of aluminium chloride. Cover and heat over a spirit lamp, at 
first gently, that the crucible be not broken by the production of 
heat inside, and then heat stronger, at last to redness. Cool, 
and when fully cold put it into a glass of cold water. A gray 
powder separates out, which on nearer observation, especially 
in sunlight, is seen to consist of little flakes of metal. After it 
has separated, pour off the solution, filter, wash with cold 
water, and dry; this is the aluminium." 

In reality, this powder possessed no metallic properties, and 
moreover, it contained potassium and aluminium chloride, 
which gave it the property of decomposing water at 100°. To 
avoid the loss of aluminium chloride by volatilization at the 
high heat developed during the reaction, Liebig afterwards 
made its vapors pass slowly over some potassium placed in a 
long glass tube. This device of Liebig is nearly the arrange- 
ment which Wohler adopted later, in 1845, ^nd which gave him 
much better results. 

Wbklcr's Experiments ( 1 845 ) . 
The following is Wohler's second paper published in 1845 ■* 

* Liebig's Annalen, 53, 422. 


" On account of the violent incandescence with which the 
reduction of aluminium chloride by potassium is accompanied, 
this operation requires great precautions, and can be carried 
out only on a small scale. I took for the operation a platinum 
tube, in which I placed aluminium chloride, and near it some 
potassium in a platinum boat. I heated the tube gently at first, 
then to redness. But the reduction may also be done by put- 
ting potassium in a small crucible, which is placed inside a 
larger one, and the space between the two filled with aluminium 
chloride. A close cover is put over the whole, and it is heated. 
Equal volumes ot potassium and the aluminium salt are the 
best proportions to employ. After cooling, the tube or cruci- 
ble is put in a vessel of water. The metal is obtained as a gray 
metallic powder, but on closer observation one can see even 
with the naked eye small tin-white globules, some as large as 
pins' heads. Under the microscope magnifying two hundred 
diameters, the whole powder resolves itself into small globules, 
several of which may sometimes be seen sticking together, 
showing that the metal was melted at the moment of reduction. 
A beaten-out globule may be again melted to a sphere in a 
bead of borax or salt of phosphorus, but rapidly oxidizes dur- 
ing the operation, and if the heat is continued, disappears 
entirely, seeming either to reduce boric acid in the borax bead, 
or phosphoric acid in the salt of phosphorus bead. I did not 
succeed in melting together the pulverulent aluminium in a 
crucibe with borax, at a temperature which would have melted 
cast-iron, for the metal disappeared entirely and the borax be- 
came a black slag. It seems probable that aluminium, being 
lighter than molten borax, swims on it and burns. The white 
metallic globules had the color and lustre of tin. It is perfectly 
malleable and can be hammered out to the thinnest leaves. 
Its specific gravity, determined with two globules weighing 32 
milligrammes, was 2.50, and with three hammered- out globules 
weighing 34 milligrammes, 2.67. On account of their lightness 
these figures can only be approximate. It is not magnetic, 
remains white in the air, decomposes water at 100°, not at 


usual temperatures, and dissolves completely in caustic potash 
(KOH). When heated in oxygen almost to melting, it is only 
superficially oxidized, but it burns like zinc in a blast-lamp 

These results of Wohler's, especially the determination of 
specific gravity, were singularly accurate when we consider 
that he established them working with microscopic bits of 
the metal, It was just such work that established Wohler's 
fame as an investigator. However, we notice that his metal 
differed from aluminium as we know it in several important 
respects, in speaking of which Deville says: "All this time 
the metal obtained by Wohler was far from being pure ; it was 
very difficultly fusible, owing, without doubt, to the fact that 
it contained platinum taken from the vessel in which it had 
been prepared. It. is well known that these two metals com- 
bine very easily at a gentle heat. Moreover, it decomposed 
water at 100°, which must be attributed either to the presence 
of potassium or to aluminium chloride, with which the metal 
might have been impregnated : for aluminium in presence of 
aluminium chloride in effect decomposes water with evolution 
of hydrogen." 

After Wohler's paper in 1845, the next improvement is that 
introduced by Deville in 1854-55, and this is really the date at 
which aluminium, the metal, became known and its true proper- 
ties established. 

Deville s Experiments (1854). 

The results of this chemist's investigations and success in 
obtaining pure aluminium were first made public at the seance 
of the French Academy, August 14, 1854, and included men- 
tion of an electrolytic method of reduction (see Chap. XI.), as 
well as of the following on reduction by sodium.* 

" The following is the best method for obtaining aluminium 
chemically pure in the laboratory. Take a large glass tube 
about four centimeters in diameter, and put into it 200 or 300 

* Ann. de Phys. et de Chim., xliii. 24. 


grammes of pure aluminium chloride free from iron, and isolate 
it between two stoppers of amianthus (fine, silky asbestos). 
Hydrogen, well dried and free from air, is brought in at one 
end of the tube. The aluminium chloride is heated in this cur- 
rent of gas by some lumps of charcoal in order to drive ofif 
hydrochloric acid or sulphides of chlorine or of silicon, with 
which it is always impregnated. Then there are introduced into 
the tube porcelain boats, as large as possible, each containing 
several grammes of sodium, which was previously rubbed quite 
dry between leaves of filter paper. The tube being full of 
hydrogen, the sodium is melted, the aluminium chloride is 
heated and distils, and decomposes in contact with the sodium 
with incandesence, the intensity of which can be moderated at 
pleasure. The operation is ended when all the sodium has dis- 
appeared, and when the sodium chloride formed has absorbed 
so much aluminium chloride as to be saturated with it. The 
aluminium which has been formed is held in the double chloride 
of sodium and aluminium, AlClg.NaCl, a compound very fusible 
and very volatile. The boats are then taken from the glass 
tube, and their entire contents put in boats made of retort car- 
bon, which have been previously heated in dry chlorine in order 
to remove all siliceous and ferruginous matter. These are then 
introduced into a large porcelain tube, furnished with a pro- 
longation and traversed by a current of hydrogen, dry and free 
from air. This tube being then heated to bright redness, the 
aluminium-sodium chloride distils without decomposition and 
condenses in the prolongation. There is found in the boats, 
after the operation, all the aluminium which had been reduced, 
collected in at most one or two small buttons. The boats when 
taken from the tube should be nearly free from aluminium- 
sodium chloride, and also from sodium chloride. The buttons 
of aluminium are united in a small earthenware crucible, which 
is heated as gently as possible, just sufificient to melt the metal. 
The latter is pressed together and skimmed clean by a small 
rod or tube of clay. The metal thus collected may be very 
suitably cast in an ingot mould." 


The later precautions added to the above given process were 
principally directed towards avoiding the attacking of the cruci- 
ble, which always takes place when the metal is melted with a 
flux, and the aluminium thereby made more or less siliceous. 
The year following the publication of these results, this labora- 
tory method was carried out on a large scale at the chemical 
works at Javel. 

Deville's Methods (1855). 

The methods about to be given are those which were devised 
in Deville's laboratory at the Ecole Normale, during the winter 
of 1854-55, and applied on a large scale at Javel, during the 
spring of 1855 (March-July). The Emperor Napoleon III. 
defrayed the expenses of this installation. Descriptions of the 
methods used for producing alumina, aluminium chloride and 
sodium at Javel can be found under their respective headings 
(pp. 134, 153, 180), We here confine our description to 
the mode of reducing the aluminium chloride by sodium, and 
the remarks incident thereto. The process has at present 
only an historic interest, as it was soon modified in its details 
so as to be almost entirely changed. The following is Deville's 
description : — 

" Perfectly pure aluminium. — To obtain aluminium perfectly 
pure it is necessary to employ materials of absolute purity, to 
reduce the metal in presence of a completely volatile flux, and 
finally never to heat, especially with a flux, in a siliceous vessel 
to a high temperature. 

" Pure materials. — The necessity of using absolutely pure 
materials is easy to understand ; all the metallic impurities are 
concentrated in the aluminium, and unfortunately I know no 
absolute method of purifying the metal. Thus, suppose we 
take an alum containing o.i per cent, of iron and 11 per cent, 
of alumina; the alumina derived from it will contain i per 
cent, of iron, and supposing the alumina to give up all the alu- 
minium in it, the metal will be contaminated with 2 per cent, 
of iron. 



'^Influence of flux or slag. — The flux, or the product of the 
reaction of the sodium on the aluminous material, ought to be 
volatile, that one may separate the aluminium by heat from the 
material with which it has been in contact, and with which it 
remains obstinately impregnated because of its small specific 

" Influence of the vessel. — The siliceous vessels in which 
aluminium is received or melted give it necessarily a large 
quantity of silicon, a very injurious impurity. Silicon cannot 
be separated from aluminium by any means, and the siliceous 
aluminium seems to have a greater tendency to take up more 
silicon than pure aluminium, so that after a small number of 
remeltings in siliceous vessels the metal becomes so impure as 
to be almost infusible. 

In order to avoid the dangers pointed out above, Deville re- 
commended following scrupulously the following details in 
order to get pure aluminium. 

" Reduction of solid sodium. — The crude aluminium chloride 
placed in the cyHnder A (Fig. 23), is vaporized by the fire and 


Fig. 24. 



I to.»»iaa Y^s;^ teMaa "T 

passes through the tube to the cylinder B, containing 60 to 80 
kilos of iron nails heated to a dull-red heat. The iron retains 
as relatively fixed ferrous chloride, the ferric chloride and 
hydrochloric acid which contaminate the aluminium chloride, 
and likewise transforms any sulphur dichloride (SCl^) in it into 
ferrous chloride and sulphide of iron. The vapors on passing 
out of B through the tube, which is kept at about 300°, deposit 


spangles of ferrous chloride, which is without sensible tension 
at that temperature. The vapors then enter D, a cast-iron 
cylinder in which are three cast-iron boats each containing 500 
grms. of sodium. It is sufficient to heat this cylinder barely to 
a dull-red heat in its lower part, for the reaction once com- 
menced disengages enough heat to complete itself, and it is 
often necessary to take awav all the fire from it. There is at 
first produced in the first boat some aluminium and some sod- 
ium chloride, which latter combines with the excess of alumin- 
ium chloride to form the volatile aluminium-sodium chloride, 
AlClj.NaCl. These vapors of double chloride condense on the 
second boat and are decomposed by the sodium into alumin- 
ium and sodium chloride. A similar reaction takes place in 
the third boat when all the sodium of the second has disap- 
peared. When on raising the cover it is seen that the 
sodium of the last boat is entirely transformed into a lumpy 
black material, and that the reactions are over, the boats 
are taken out, immediately replaced by others, and are al- 
lowed to cool covered by empty boats. In this first opera- 
tion the reaction is rarely complete, for the sodium is 
protected by the layer of sodium chloride formed at its ex- 
pense. To make this disappear, the contents of the boats are 
put into cast-iron pots or earthen crucibles, which are heated 
until the aluminium chloride begins to volatilize, when the so- 
dium will be entirely absorbed and the aluminium finally re- 
mains in contact with a large excess of its chloride, which is 
indispensable for the success of the operation. Then the pots 
or crucibles are cooled, and there is taken from the upper part 
of their contents a layer of sodium chloride almost pure, while 
underneath are found globules of aluminium which are sepa- 
rated from the residue by washing with water. Unfortunately 
the water in dissolving the aluminium chloride of the flux exer- 
cises on the metal a very rapid destructive action, and only 
the globules larger than the head of a pin are saved from this 
washing. These are gathered together, dried, melted in an 
earthen crucible, and pressed together with a clay rod. 


The button is then cast in an ingot mould. It is important in 
this operation to employ only well-purified sodium, and not to 
melt the aluminium if it still contains any sodium, for in this 
case the metal takes fire and burns as long as any of the alka- 
line metal remains in it. In such a case it is necessary to re- 
melt in presence of a little aluminium-sodium chloride. 

" Such is the detestable process by means of which were made 
the ingots of aluminium sent to the Exhibition (1855). To 
complete my dissatisfaction at the processs, pressed by time and 
ignorant of the action of copper on aluminium, I employed in 
almost all my experiments reaction cylinders and boats of cop- 
per, so that the aluminium I took from them contained such 
quantities of this metal as to form a veritable alloy. Moreover, it 
had lost almost all ductility and malleability, had a disagreeable 
gray tint, and finally at the end of two months it tarnished by 
becoming covered with a black layer of oxide or sulphide of 
copper, which could only be removed by dipping in nitric acid. 
But, singular to relate, an ingot of virgin silver which had been 
put alongside the aluminium that the public might note easily 
the difference in color and weight of the two metals, was 
blackened still worse than the impure aluminium. Only one of 
the bars exhibited, which contained no copper, remained un- 
altered from the day of its manufacture till now (1859). It 
was some of this cupreous aluminium that I sent to M. 
Regnault, who had asked me for some in order to determine its 
specific heat. I had cautioned him at the time of the number 
and nature of the impurities which it might contain, and the 
analysis of M. Salvetat, which is cited in the memoir of 
Regnault, accords with the mean composition of the specimens 
that I had produced and analyzed at that time (see p. 54, 
Analysis i). It is to be regretted that I gave such impure 
material to serve for determinations of such splendid precision. 
I was persuaded to do so only by the entreaties of M. 
Regnault, who could not wait until I prepared him better. It is 
also this cupreous aluminium which M. Hulot has called ' hard 
aluminium,' in a note on the physical properties of this metal 


which he addressed to the Academy. Hulot has remarked 
that this impure metal, which is crystalline in structure, after 
having been compressed between the dies of the coining press 
may lose its crystalline structure, to which it owes its brittle- 
ness, and become very malleable. It possesses then such 
strength that it works well in the rolls of a steel rolling mill. 
Further, this 'hard aluminium' becomes quite unalterable when 
it has thus lost its texture. 

"Reduction by sodium vapor. — This process, which I have not 
perfected, is very easy to operate, and gave me very pure metal 
at the first attempt. I operate as follows : I fill a mercury bot- 
tle with a mixture of chalk, carbon, and carbonate of soda, in 
the proportions best for generating sodium. An iron tube 
about ten centimetres long is screwed to the bottle, and the 
whole placed in a wind furnace, so that the bottle is heated to 
red-white and the tube is red to its end. The end of the tube 
is then introduced into a hole made in a large earthen crucible 
about one-fourth way from the bottom, so that the end of the 
tube just reaches the inside surface of the crucible. The car- 
bonic oxide (CO) disengaged burns in the bottom of the cru- 
cible, heating and drying it ; afterwards the sodium flame 
appears, and then pieces of aluminium chloride are thrown into 
the crucible from time to time. The salt volatilizes and de- 
composes before this sort of tuyere, from which issues the 
reducing vapor. More aluminium salt is added when the 
vapors coming from the crucible cease to be acid, and when 
the flame of sodium burning in the atmosphere of aluminium 
chloride loses its brightness. When the operation is finished, 
the crucible is broken and there is taken from the walls below 
the entrance of the tube a saline mass composed of sodium 
chloride, a considerable quantity of globules of aluminium, and 
some sodium carbonate, which latter is in larger quantity the 
slower the operation was performed. The globules are de- 
tached by plunging the saline mass into water, when it be- 
comes necessary to notice the reaction of the water on litmus. 
If the water becomes acid, it is renewed often ; if alkaline, the 


mass, impregnated with metal, must be digested in nitric acid, 
diluted with three or four volumes of water, and so the metal is 
left intact. The globules are reunited by melting with the 
precautions before given." 

Deville's Process (1859). 

The process then in use at Nanterre was based on the use of 
aluminium-sodium chloride, which was reduced by sodium, 
with cryolite or fluorspar as a flux. The methods of preparing 
each of these materials were carefully studied out at the chem- 
ical works of La Glaciere, where, from April, 1856, to April, 
1857, the manufacture of aluminium was carried on by a com- 
pany formed by Deville and a few friends, and from thence 
proceeded the actual system which was established at Nanterre 
under the direction of M. Paul Morin when the works at La 
Glaciere were closed. The methods of preparation of alumina, 
aluminium-sodium chloride and sodium used at Nanterre are 
placed under their respective, headings (pp. 137, 154, 190). 

In the first year that the works at Nanterre were in opera- 
tion there were made : 

Aluminium-sodium chloride 10,000 kilos. 

Sodium 2,000 " 

Aluminium 600 " 

The metal prepared improved constantly in quality, M. Morin 
profiting continually by his experience and improving the 
practical details constantly, so that the aluminium averaged, in 
1859, gj per cent. pure. (See Analysis 9, p. 54.) 

As to the rationale of the process used, aluminium chloride 
was replaced by aluminium-sodium chloride because the latter 
is less deliquescent and less difficult of preservation ; but the. 
small amount of moisture absorbed by the double chloride is 
held very energetically, at a high temperature giving rise to 
some alumina, which encloses the globules of metal with a thin 
coating, and so hinders their easy reunion to a button. Deville 
remarked that the presence of fluorides facilitated the reunion 


of these globules, which fact he attributed to their dissolving 
the thin coat of alumina ; so that the employment of a fluoride 
as a flux became necessary to overcome the effect produced 
primarily by the aluminium-sodium chloride holding moisture 
so energetically. Deville gives the following account of the 
development of these improvements : "The facility with which 
aluminium unites in fluorides is due without doubt to the prop- 
erty which these possess of dissolving the alumina on the sur- 
face of the globules at the moment of their formation, and 
which the sodium is unable to reduce. I had experienced great 
difificulty by obtaining small quantities of metal poorly united, 
when I reduced the aluminium-sodium chloride by sodium ; 
M. Rammelsberg, who often made the same attempts, tells me 
he has had a Hke experience. But I am assured by a scrupu- 
lous analysis that the quantity of metal reduced by the sodium 
is exactly that which theory indicates, although after many 
operations there is found only a gray powder, resolving itself 
under the microscope into a multitude of small globules. The 
fact is simply that aluminium-sodium chloride is a very poor 
flux for aluminium. MM. Morin, Debray, and myself have 
undertaken to correct this bad effect by the introduction of a 
solvent for the alumina into the saline slag which accompanies 
the aluminium at the moment of its formation. At first, we 
found it an improvement to condense the vapors of aluminium 
chloride, previously purified by iron, directly in sodium chloride, 
placed for this purpose in a crucible and kept at a red heat. 
We produced in this way, from highly colored material, a double 
chloride very white and free from moisture, and furnishing on 
reduction a metal of fine appearance. We then introduced 
fluorspar (CaF^) into the composition of the mixture to be 
reduced, and we obtained good results with the following pro- 
portions : 

Aluminium-sodium chloride 400 grammes. 

Sodium chloride 200 " 

Calcium fluoride 200 " 

Sodium 75 to 80 " 


" The double chloride ought to be melted and heated almost 
to low red heat at the moment it is employed, the sodium 
chloride calcined and at a red heat or melted, and the fluorspar 
pulverized and well dried. The double chldride, sodium chlor- 
ide and calcium fluoride are mixed and alternated in layers in 
the crucible with sodium. The top layer is of the mixture, and 
the cover is sodium chloride. Heat gently, at first, until the 
reaction ends, and then to a heat about sufficient to melt silver. 
The crucible, or at least that part of it which contains the mix- 
ture, ought to be of a uniform red tint, and the material per- 
fectly liquid. It is stirred a long time and cast on a well dried, 
chalked plate. There flows out first a very limpid liquid, col- 
orless and very fluid ; then a gray material, a little more pasty, 
which contains aluminium in little grains, and is set aside ; and 
finally a button with small, metallic masses which of them- 
selves ought to weigh 20 grms. if the operation has succeeded 
well. On pulverizing and sieving the gray slag, 5 or 6 grms. 
of small globules are obtained, which may be pressed together 
by an earthen rod in an ordinary crucible heated to redness. 
The globules are thus reunited, and when a sufficient quantity 
is collected the metal is cast into ingots. In a well-conducted 
operation, 75 grms. of sodium ought to give a button of 20 
grms. and S grms. in grains, making a return of one part alu- 
minium from three of sodium. Theory indicates one to two 
and a half, or 30 grms. of aluminium from 75 of sodium. But 
all the efforts which have been made to recover from the inso- 
luble slag the 4 or 5 grms. of metal not united, but easily vis- 
ible with a glass, have been so far unsuccessful. There is, 
without doubt, a knack, a particular manipulation, on which de- 
pends the success of an operation which would render the 
theoretical amount of metal, but we lack it yet. These opera- 
tions take place, in general, with more facility on a large scale, 
so that we may consider fluorspar as being suitable for serving 
in the manufacture of aluminium in crucibles. We employed 
very pure fluorspar, and our metal was quite exempt from 
silicon. It is true that we took a precaution which is necessary 


to adopt in operations of this kind ; we plastered our crucibles 
inside with a layer of aluminous paste, the composition of 
which has been given in 'Ann. de Chim. et de Phys.' xlvi. 195. 
This paste is made of calcined alumina and an aluminate of 
lime, the latter obtained by heating together equal parts of 
chalk and alumina to a high heat. By taking about four parts 
calcined alumina and one of aluminate of lime well pulverized 
and sieved, moistening with a little water, there is obtained a 
paste with which the inside of an earthen crucible is quickly 
and easily coated. The paste is spread evenly with a porcelain 
spatula, and compressed strongly until its surface has become 
well polished. It is allowed to dry, and then heated to bright 
redness to season the coating, which does not melt, and pro- 
tects the crucible completely against the action of the alumin- 
ium and fluorspar. A crucible will serve several times in suc- 
cession, provided that the new material is put in as soon as the 
previous charge is cast. The advantages of doing this are 
that the mixture and the sodium are put into a crucible al- 
ready heated up, and so lose less by volatilization because the 
heating is done more quickly, and the crucible is drier than if 
a new one had been used or than if it had been let cool. A 
new crucible should be heated to at least 300° or 400° before 
being used. The saline slag contains a large quantity of cal- 
cium chloride, which can be washed away by water, and an in- 
soluble material from which aluminium fluoride can be volatil- 

" Yet the operation just described, which was a great im- 
provement on previous ones, requires many precautions and a 
certain skill of manipulation to succeed every time. But 
nothing is more easy or simple than to substitute cryolite for 
the fluorspar. Then the operation is much easier. The 
amount of metal produced is not much larger, although the 
button often weighs 22 grammes, yet if cryolite can only be ob- 
tained in abundance in a continuous supply, the process which 
I will describe will become most economical. The charge is 
made up as before, except introducing cryolite for fluorspar. 


In one of our operations we obtained, with -j^ grms. of sodium, 
a button weighing 22 grms. and 4 grms. in globules, giving a 
yield of one part aluminium to two and eight-tenths parts 
sodium, which is very near to that indicated by theory. The 
metal obtained was of excellent quality. However, it con- 
tained a little iron coming from the aluminium chloride, which 
had not been purified perfectly. But iron does not injure the 
properties of the metal as copper does ; and, save a little bluish 
coloration, it does not alter its appearance or its resistance to 
physical and chemical agencies. 

" After these attempts we tried performing the reduction sim- 
ply on the bed of a reverberatory furnace, relying on the imme- 
diate reaction of sodium on the double chloride to use up these 
materials before they could be perceptibly wasted by the fur- 
nace gases. This condition was realized in practice with 
unlooked-for success. The reduction is now made on a some- 
what considerable scale at the Nanterre works, and never, since 
commencing to operate in this way, has a reduction failed, the 
results obtained being always uniform. The furnace now used 
has all the relative dimensions of a soda furnace. In fact, 
almost the temperature of an ordinary soda furnace is required 
that the operation may succeed perfectly. The absolute dimen- 
sions of the furnace, however, may vary with the quantity of alu- 
minium to be made in one operation, and are not limited. With 
a bed of one square metre surface, 6 to 10 kilos of aluminium 
can be reduced at once ; and since each operation lasts about 
four hours, and the furnace may be recharged immediately after 
emptying it of the materials just treated, it is seen that, with so 
small a bed, 60 to lOO kilos of aluminium can be made in 
twenty-four hours without any difficulty. In this respect, I 
think the industrial problem perfectly solved. The proportions 
which we employed at first were — 

Aluminium-sodium chloride (crushed) 10 parts. 

Fluorspar 5 " 

Sodium (in ingots) 2 " 

" As aluminium is still very dear, it is necessary to direct 


great attention to the return from the materials used, and on 
this point there is yet much progress to be made. We ascer- 
tained many times that the return was always a little better, and 
the reunion of the metal to a single ingot a little easier, when 
cryolite was substituted for fluorspar, the price of the former, 
after having been high, being now lowered to 350 francs per 
tonne (about $75 per long ton). For this reason we can now 
use cryolite instead of fluorspar, and in the same proportions. 
We can also recover alumina from this cryolite by treating the 
slags (see p. 148). The double chloride and pulverized cryo- 
lite are now mixed with the sodium in small ingots, and the 
mixture thrown on to the bed of the heated-up furnace. The 
dampers are then shut, to prevent as much as possible access 
of air. Very soon a lively reaction begins, with the production 
of such heat that the brick sides of the furnace, as well as the 
materials on the hearth, are made bright red-hot. At this heat 
the mixture is almost completely fused. Then it is necessary 
to open the damper and direct the flame on the bed in such 
manner as to heat the bath equally all over and unite the re- 
duced aluminium. When the operation is considered ended, a 
casting is made by an opening in the back of the furnace and 
the slag is received in cast-iron pots. At the end of the cast 
the aluminium arrives in a single jet, which unites into a single 
lump at the bottom of the still-liquid slag. The gray slag flow- 
ing out last should be pulverized and sieved, to extract the 
divided globules of aluminium, 200 to 300 grammes of which 
can sometimes be extracted from one kilo of gray slag. The 
pulverization of the slag is in all cases indispensable in its sub- 
sequent treatment for extracting its alumina. The slag is of 
two kinds, one fluid and light, which covered the bath, and is 
rich in sodium chloride ; the other less fusible and pasty, gray 
in color, which is more dense and lies in contact with the alu- 
minium. The coloring material producing the grayness is car- 
bon, coming either from the sodium or from the oil which im- 
pregnated it, or finally from the vapor of the oil. I attribute 
the slight pastiness of this slag to a little alumina dissolved by 
the fluorides. This slag contains about: 


Sodium chloride 60 parts. 

Aluminium fluoride 40 " 

and on washing it the former dissolves while the latter remains, 
mixed with a little cryolite or alumina. This is the alumina 
which had been dissolved or retained in the bath of fluoride. 
It will be remarked that the bath of slag contains no other 
fluoride than aluminium fluoride, which does not attack earthen 
crucibles or siliceous materials in general except at a very high 
temperature. It is for this reason that the hearth and other parts 
of the furnace resist easily a fluoride slag containing only alu- 
minium fluoride, which has not the property of combining with 
silicon fluoride at the expense of the silica of the bricks — as 
sodium fluoride does in like circumstances. In our operations, 
cryolite is used only as a flux. In the process of reduction 
based on cryolite alone, the sodium fluoride resulting is, on the 
contrary, very dangerous to crucibles, and it is due to that fact 
especially that the aluminium absorbs a large quantity of 
silicon, which always happens with this method. In fact, it is 
well known that metallic silicon can be prepared in this way by 
prolonging the operation a little." 

Deville closes his account of the aluminium industry in 1859 
with these words : " Many things yet remain to us to do, and 
we can scarcely say now that we know the true qualities of the 
substances we employ. But the matter is so new, is harassed 
with so many difiSculties even after all that has been done, that 
our young enterprise may hope everything from the future when 
it shall have acquired experience. I ought to say, however, 
that the aluminium industry is now at such a point that if the 
uses of the metal are rapidly extended it may change its aspect 
with great rapidity. One may ask to-day how much a kilo of 
iron would cost if a works made only 60 to 100 kilos of it a 
month, if large apparatus were excluded from this industry, and 
iron obtained by laboratory processes which would permit it to 
become useful only by tedious after-treatment. Such will not 
be the case with aluminium, at least with the processes just de- 
scribed. In fact in all I undertook, either alone or with my 


friends, I have always been guided by this thought — that we 
ought to adopt only such apparatus as is susceptible of being 
immediately enlarged, and to use only materials almost as com- 
mon as clay itself for the source of the aluminium." 

The Deville Process (1882). 

The process just described reached a fair degree of perfec- 
tion at Nanterre, under the direction of M. Paul Morin. After- 
wards, some of the chemical operations incidental to the 
process were carried on at the works of the Chemical Manufac- 
turing Company of Alais and Carmargue, at Salindres (Gard), 
owned by H. Merle & Co. At a later date the whole manufac- 
ture was removed to this place, while the Societe Anonyme de 
1' Aluminium, at Nanterre, worked up the metal, and placed it 
on the market. The Salindres works, about 1880, went under 
the management of A. R. Pechiney & Co., and under the per- 
sonal attention of M. Pechiney, the Deville process reached its 
perfection. The following account is taken mostly from M. 
Margottet's article on aluminium in Fremy's Encyclopedic 

An outline of the process, as it was prepared, may very 
appropriately be given at this place, although detailed descrip- 
tions of the preliminary processes for preparing the materials 
for reduction are given under the appropriate headings (see 
pp. 137, 154, 190). 

The primary material to furnish the aluminium is bauxite. 
To obtain the metal it is necessary to proceed successively 
through the following operations: — 

I. Preparation of the aluminate of soda, and solution of this 
salt to separate it from the ferric oxide contained in the 

II. Precipitation of hydrated alumina from the aluminate of 
soda by a current of carbon dioxide ; washing the precipitate. 

III. Preparation of a mixture of alumina, carbon, and salt, 
drying it, and then treating with gaseous chlorine to obtain the 
double chloride of aluminium and sodium. 


IV. Lastly, treatment of this chloride by sodium to obtain 

The principal chemical reactions on which this process rests 
are the following: — 

Formation of aluminate of soda by calcining bauxite with 
sodium carbonate — 

(AlFe) A.2H,0 + 3Na,C03=-Al,03-3Na,0 + Fe.Oa 
+ 2H,0 + 3CO,. 

Formation of alumina by precipitating the aluminate of soda 
with a current of carbon dioxide — 

Al,03-3Na,0 + 3CO, + 3H,0 = Al.Os.sH.O + 3Na,C03. 

Formation of aluminium-sodium chloride by the action of 
chlorine on a mixture of alumina, carbon, and sodium chloride — 

AI2O3 + 3C + 2NaCl + 6C1 = 2 (AlCla.NaCl) + 3CO. 

Reduction of this double chloride by sodium — 

AlCIs.NaCl + 3Na = Al + 4NaCl. 

As observed before, we will here consider only the last oper- 
ation. The advances made since 1859 are mostly in matters of 
detail, which every one knows are generally the most import- 
ant part of a process ; and so, although a few of the details 
may be repeated, yet we think it best not to break the contin- 
uity of this description by excising those few sentences which 
are nearly identical in the two accounts. 

The difiSculty of this operation, at least from an industrial 
point of view, is to get a slag fusible enough and light enough 
to let the reduced metal easily sink through it and unite. 
This result has been reached by using cryolite. This material 
iorms with the sodium chloride resulting from the reaction a 
very fusible slag, in the midst of which the aluminium collects 
well, and falls to the bottom. In one operation the charge is 
now composed of — 


100 kilos Aluminium-sodium chloride. 

45 " Cryolite. 

35 " Sodium. 

The double chloride and cryolite are pulverized, the sodium^ 
cut into small pieces a little larger than the thumb, is divided 
into three equal parts, each part being put into a sheet-iron 
basket. The mixture of double chloride and cryolite, being 
pulverized, is divided into four equal parts, three of these are 
respectively put in each basket with the sodium, the fourth 
being placed in a basket by itself. The reduction furnace (see 

Fig. 25. 



Fig. 25) is a little furnace of refractory brick, with an inclined 
hearth and a vaulted roof. This furnace is strongly braced by 
iron tie-rods, because of the concussions caused by the reac- 
tion. The flame may at any given moment be directed into a. 
flue outside of the hearth. At the back part of the furnace,, 
that is to say, on that side towards which the bed slopes, is a 
little brick wah which is built up for each reduction and is. 
taken away in operating the running out of the metal and slag> 
A gutter of cast-iron is placed immediately in front of the wall 
to facilitate running out the materials. All this side of the fur- 
nace ought to be opened or closed at pleasure by means of a 
damper. Lastly, there is an opening for charging in the roof, 
closed by a lid. At the time of an operation the furnace 
should be heated to low redness, then are introduced in rapid 


succession the contents of the three baskets containing sodium, 
etc., and lastly the fourth containing only double chloride and 
no sodium. Then all the openings of the furnace are closed, 
and a very vivid reaction accompanied by dull concussions im- 
mediately takes place. At the end of fifteen minutes, the ac- 
tion subsides, the dampers are opened and the heat continued, 
meanwhile stirring the mass from time to time with an iron 
poker. At the end of three hours the reduction is ended, and 
the metal collects at the bottom of the liquid bath. Then the 
running out is proceeded with in three phases : First — Running 
off the upper part of the bath, which consists of a fluid material 
completely free from reduced aluminium and constituting the 
white slag. To run this out a brick is taken away from the 
upper course of the little wall which terminates the hearth. 
These slags are received in an iron wagon. Second — Running 
out the aluminium. This is done by opening a small orifice 
left in the bottom of the brick wall, which was temporarily 
plugged up. The liquid metal is received in a cast-iron melt- 
ing pot, the bottom of which has been previously heated to 
redness. The aluminium is immediately cast in a series of 
small rectangular cast-iron moulds. Third — Running out of 
the rest of the bath, which constitutes the gray slags. These 
were, like the white slags, formed by the sodium chloride and 
cryolite, but they contain, in addition, isolated globules of alu- 
minium. To run these out, all the bricks of the little wall are 
taken away.. , This slag is received in the same melting pot into 
which the aluminium was run, the latter having been already 
moulded. Here it cools gradually, and after cooling there are 
always found at the bottom of the pot several grains of metal. 
In a good operation there are taken from one casting 10.5 
kilos of aluminium, which is sold directly as commercial metal. 

The following data as to the expense of this process may be 
very appropriately inserted here, giving the cost at Salindres in 
1873, in which year 3600 kilos are said to have been made. 

* Manufacture of one kilo of aluminium. 

* A. Wurtz, Wagner's Jahresb., 1874, vol. xxi. 


Sodium 3.44 kilos @ 11.32 fr. per kilo = 38 fr. 90 cent. 


chloride. . . . 10.04 " 2.48 " " = 24 " 90 " 

Cryolite 3.87 " 61.0 " 100 kilos ^ 2 " 36 " 

Coal 29.17 " 1.40 " " = o " 41 " 

Wages I " 80 " 

Costs o " 88 " 

Total 69 " 25 " 

This must be increased ten per cent, for losses and other ex- 
penses, making the cost of aluminium 80 fr. per kilo, and it is 
sold for 100. ($9.00 per lb.) 

According to a statement in the ' Bull, de la Soc. de I'lndustrie 
Minerale,' ii. 451, made in 1882, Salindres was then the only 
place in which aluminium was being manufactured. The 
Deville process was finally driven out of the market, and the 
manufacture at Salindres suspended, in 1890. 

Niewerth's Process (1883). 

This method can be regarded as little more than a suggestion, 
since it follows exactly the lines of some of Deville's earlier ex- 
periments. Although theoretically very advantageous, yet in 
practice it has probably been found far inferior in point of yield 
of metal and expense to the ordinary sodium processes. The 
patent is taken out in the United States and other countries in 
the name of H. Niewerth, of Hanover, and is thus summarized : 

*A compound of aluminium, with chlorine or fluorine, is 
brought by any means into the form of vapor, and conducted, 
strongly heated, into contact with a mixture of 62 parts sodium 
carbonate, 28 coal, and 10 chalk, which is also in a highly heated 
condition. This mixture disengages sodium, which reduces the 
gaseous chloride or fluoride of aluminium, the nascent sodium 
being the reducing agent. In place of the above mixture other 
suitable mixtures which generate sodium may be employed, or 
mixtures may also advantageously be used from which potassium 
is generated. 

* Sci. Am. Supple., Nov. 17, 1883. 


Gadsden's Patent (1883). 

H. A. Gadsden, of London, and E. Foote,* of New York, 
were granted a patent based on the principle of heating in a 
retort sodium carbonate and carbonaceous matter, or any suit- 
able mixture for generating sodium, and conducting the vapor 
of sodium produced into another retort, lined with carbon, in 
which aluminium chloride, or aluminium-sodium chloride or 
cryolite, has been placed and heated. The second English pat- 
ent claims to heat a mixture which will generate sodium in one 
retort, and pass chlorine over a mixture of carbon and alum- 
ina, thus generating aluminium chloride, in another retort, and 
then mixing the two vapors in a third retort or reaction 

Friskmuth's Process (1884). 

This was patented in the United States in 1884 (U. S. Pat. 
308,152, Nov. 1884). In what the originality of the process 
consists, in view of Deville's publications and even in view of 
the processes just mentioned, we cannot see. Col. Frishmuth 
himself admitted, in 1887, having abandoned the sodium process, 
as the difficulties of the method did not permit its competing 
with the more roundabout but more easily-conducted operation 
with solid sodium. A simple transcript of the claims in his 
patent will give a sufficiently extended idea of the reactions 
proposed to be used. 

1. The simultaneous generation of sodium vapor and a vola- 
tile compound of aluminium in two separate vessels or retorts, 
and mingling the vapors thus obtained in a nascent ( ! ) state 
in a third vessel, wherein they react on each other. 

2. The sodium vapor is produced from a mixture of a sodium 
compound and carbon, or some other reducing agent; and the 
aluminous vapor from aluminous material. 

3. The simultaneous generation of sodium vapor and vapor 
of aluminium chloride or aluminium fluoride; or of sodium 
vapor and aluminium-sodium chloride. 

* English patents 1995 and 493° (1883); German patent 27,572 (1884). 


4. Converting the aluminous material to a vapor by heating 
it in a retort with sodium chloride, and subjecting it at the 
same time to chlorine gas ; mingling the vapor of aluminium- 
sodium chloride thus obtained with vapor simultaneously gen- 
erated from sodium carbonate and carbon. 

H. von Grousillier' s Improvement (1885)- 

This suggestion as to the way of performing the reduction 
by sodium is the subject of the English patent 7858, June 29, 
18B5. Dr. Fischer remarks on it, in " Wagner's Jahresbericht" 
for 1885, that "it is apparently wholly worked out at the 
writing-table," The patentee. Hector von Grousillier, Springe, 
Hanover, thus describes his invention : 

" In order to avoid the difficulties ordinarily met with in the 
use of aluminium-sodium chloride to obtain aluminium, I raise 
the volatilizing point of aluminium chloride by performing its 
reduction, either chemically or electrolytically, under pressure 
in a strong hermetically-closed vessel, lined with clay or mag- 
nesia, and provided with a safety valve." 

The Deville-Castner Process (1886). 

This last development of the old Deville process was ope- 
rated by the Aluminium Company, Limited, at their large new 
works at Oldbury, near Birmingham, England. The plant 
covered nearly five acres of ground, and adjoined Chance Bros.' 
large chemical works, from which the hydrochloric acid used 
was obtained and the waste soda-liquors returned, by means of 
large pipes connecting the two plants. The company was thus 
in position to obtain acid and dispose of its by-products to 
very good advantage. The principle on which the process was 
worked is similar to its predecessor, in being the reduction of 
aluminium-sodium chloride by sodium, but it improves on the 
other in the cheaper production of both these materials. For 
instance, the alumina used is obtained and converted into 
double chloride by Mr. Webster's processes, by which the cost 
of this salt is not over 3^. per lb. (see p. 164), as against i2d.. 


the cost at Salindres ; further, by Mr. Castner's sodium process 
it is acknowledged that the sodium cost only about ^d. per lb., 
as against 48^^., or $1, as formerly. Since 10 lbs. of the chlo- 
ride and 3 lbs. of sodium are required to produce i lb. of alu- 
minium, the average saving in these two items, over the old 
process, is somewhere about 75 per cent. 

The works contain a sodium building, in which are four large 
sodium furnaces, each capable of producing over 500 lbs. of 
that metal in twenty-four hours ; the sodium is also remelted 
and stored in the same building (see p. 215). The double 
chloride furnaces are in a building 250 feet by 50 feet wide, there 
being 12 furnaces, each containing 5 retorts. The total output 
of double chloride is an average of 5000 lbs. per day. (See. p. 
162). Connected with this building is a chlorine plant of the 
largest size, capable of supplying about a ton and a half of 
chlorine a day. In a separate building are two reverberatory 
furnaces, in which the final reduction takes place and the alu- 
minium is produced. Besides these there are a rolling mill, 
wire mill, and foundry on the grounds. From the quantity of 
sodium and double chloride produced, we can see that the 
works can produce about 500 lbs. of aluminium a day, or 
150,000 lbs. a year, with some sodium left over for sale or other 

The mode of conducting the reduction is not very different 
from that practiced at Salindres. There are two regenerative 
reverberatory furnaces used, one about twice as large as the 
other. The larger furnace has a bed about six feet square, slop- 
ing towards the front of the furnace, through which are several 
openings at different heights. The charge for this furnace con- 
sists of 1200 lbs. of double chloride, 350 lbs. of sodium, and 
600 lbs. of cryolite for a flux. The chloride is in small pieces, 
the cryolite is in powder, and the sodium is cut into thin slices 
by a machine. These ingredients are put into a revolving 
wooden drum placed on a staging over the furnace, and are 
there thoroughly mixed. The drum is then opened and 
turned, when the contents fall into a small wagon beneath. 


The furnace having been raised to the required temperature, all 
the dampers are shut and the car is moved on a track immedi- 
ately over a. large hopper placed in the roof of the furnace. 
The hopper being opened, the charge is dumped in and drops 
on to the centre of the hearth. The reaction is immediate, 
and the whole charge becomes liquid in a very short time. 
After a few minutes, heating gas is again turned on, and the 
furnace kept moderately hot for two or three hours. The reac- 
tion has been 

AlCla.NaCl + 3Na = Al + 4NaCl 

and the aluminium gathers under the bath of cryolite and sod- 
ium chloride. One of the lower tap-holes is then opened with 
a bar, and the aluminium run out into moulds. When the 
metal has all run out it is followed by slag, which flows into 
iron wagons. The openings are then plugged up and the fur- 
nace is ready for another charge. The charge given produces 
usually 115 to 120 lbs. of aluminium, the whole operation last- 
ing about 4 hours. The large furnace could thus produce 840- 
lbs. in 24 hours, and the smaller one half that quantity. The 
first portion of metal running out is the purest, the latter por- 
tions, and especially that entangled in the slag on the hearth, and 
which has to be scraped out, containing more foreign sub- 
stances. This impure metal is about one-fourth of all the 
aluminium in the charge. 

The purity of the metal run out depends directly on the 
purity of the chloride used. If the double chloride contains 
0.2 per cent, of iron, the metal produced will very probably 
contain all of it, or 2 per cent. Using the double chloride 
purified by Mr. Castner's new method (see p. 163), by which 
the content of iron is reduced to 0.05 per cent, or less, alu- 
minium can be made containing less than 0.5 per cent, of iron 
and from 99 to 99.5 per cent, of aluminium. Professor Roscoe 
exhibited at one of his lectures a mass of metal weighing 116' 
lbs., being one single running from the furnace, and which con- 
tained only 0.3 per cent, of silicon and 0.5 per cent. iron. In. 


practice, the metal from 8 or 10 runnings is melted down to- 
gether to make a uniform quality. 

Taking the figures given, it appears that the metal run out 
represents 70 per cent, of the aluminium in the charge, and 80 
percent, of the weight which the sodium put in should reduce; 
but since an indeterminate weight is sifted and picked from the 
slag, it is probable that the utilization of the materials is more 
perfect than the above percentages. However, this seems to 
be the part of the old Deville process least improved upon in 
these new works, for there seems to be plenty of room for im- 
provement in perfecting the utilization of materials, especially 
in regard to loss of sodium by volatilization, which undoubt- 
edly takes place and which can possibly be altogether pre- 

The above description, written in 1890, was followed only 
one year later by the closing of this splendid works. The very 
lowest point to which the Deville-Castner process could reduce 
the cost of producing aluminium was between 4 and S shillings 
per pound ; and when the metal was put on the market by users 
of the electrolytic process at $1.50 per pound, in 1891, there 
was nothing left for the sodium process except to give up the 
business. The company, however, continues to run the works, 
making and selling large quantities of sodium. 




Methods based on the reduction of cryolite. 
These can be most conveniently presented in chronological 

Rose's Experiments (1855). 

We will here give H. Rose's entire paper, as an account of 
this eminent chemist's investigations written out by himself 
with great detail, describing failures as well as successes, 
cannot but be of value to all interested in the production of 

" Since the discovery of aluminium by Wohler, Deville has 
recently devised the means of procuring the metal in large, solid 
masses, in which condition it exhibits properties with which we 
were previously unacquainted in its more pulverulent form as 
procured by Wohler's method. While, for instance, in the lat- 
ter state it burns vividly to white earthy alumina on being ig- 
nited, the fused globules may be heated to redness without 
perceptibly oxidizing. These differences may be ascribed to 
the greater amount of division on the one hand and of density 
on the other. According to Deville, however, Wohler's metal 
contains platinum, by which he explains its difficulty of fusion, 
although it affords white alumina by combustion. Upon the 
publication of Deville's researches, I also tried to produce 
aluminium by the decomposition of aluminium-sodium chloride 
by means of sodium. I did not, however, obtain satisfactory 

* Pogg. Annalen, Sept., 1855. 


results. Moreover, Prof. Rammelsberg, who followed exactly 
the method of Deville, obtained but a very small product, and 
found it very difficult to prevent the cracking of the glass tube 
in which the experiment was conducted, by the action of the 
vapor of sodium on aluminium chloride. It appeared to me 
that a great amount of time, trouble and expense, as well as 
long practice, was necessary to obtain even small quantities of 
this remarkable metal. 

" The employment of aluminium chloride and its compounds 
with alkali chlorides is particularly inconvenient, owing to their 
volatility, deliquescence, and to the necessity of preventing all 
access of air during their treatment with sodium. It very soon 
occurred to me that it would be better to use the fluoride of 
aluminium instead of the chloride ; or rather the combination 
of the fluoride with alkaline fluorides, such as we know them 
through the investigations of Berzelius, who pointed out the 
strong affinity of aluminium fluoride for sodium fluoride and 
potassium fluoride, and that the mineral occurring in nature 
under the name of cryolite was a pure compound of aluminium 
fluoride and sodium fluoride. 

" This compound is as well fitted for the preparation of 
aluminium by means of sodium as aluminium chloride or 
aluminium-sodium chloride. Moreover, as cryolite is not vola- 
tile, is readily reduced to the most minute state of division, is 
free from water and does not attract moisture from the air, it 
affords peculiar advantages over the above-mentioned com- 
pounds. In fact, I succeeded with much less trouble in pre- 
paring aluminium by exposing cryolite together with sodium 
to a strong red heat in an iron crucible, than by using alu- 
minium chloride and its compounds. But the scarcity of cryo- 
lite prevented my pursuing the experiments. In consequence 
of receiving, however, from Prof. Krantz, of Bonn, a consider- 
able quantity of the purest cryolite at a very moderate price 
($2 per kilo), I was enabled to renew the investigation. 

" I was particularly stimulated by finding, most unexpect- 
edly, that cryolite was to be obtained here in Berlin commer- 


cially at an inconceivably low price. Prof. Krantz had already 
informed me that cryolite occurred in commerce in bulk, but 
could not learn where. Shortly after, M. Rudel, the manager 
of the chemical works of H. Kunheim, gave me a sample of a 
coarse white powder, large quantities of which were brought 
from Greenland by way of Copenhagen to Stettin, under the 
name of mineral soda, and at the price of $3 per centner. 
Samples had been sent to the soap boilers, and a soda-lye 
had been extracted from it by means of quicklime, especially 
adapted to the preparation of many kinds of soap, probably from 
its containing alumina. It is a fact, that powdered cryolite is 
completely decomposed by quicklime and water. The fluoride 
of lime formed contains no alumina, which is all dissolved by the 
caustic soda solution ; and this, on its side, is free from fluor- 
ine, or only contains a minute trace. I found this powder to 
be of equal purity to that received from Prof. Krantz. It dis- 
solved without residue in hydrochloric acid (in platinum ves- 
sels) ; the solution evaporated to dryness with sulphuric acid, 
and heated till excess of acid was dissipated, gave a residue 
which dissolved completely in water, with the aid of a little 
hydrochloric acid. From this solution, ammonia precipitated 
a considerable quantity of alumina. The solution filtered from 
the precipitate furnished, on evaporation, a residue of sulphate 
of soda, free from potash. Moreover, the powder gave the 
well-known reactions of fluorine in a marked degree. This 
powder was cryolite of great purity : therefore the coarse pow- 
der I first obtained was not the form in which it was originally 
produced. It is now obtainable in Berlin in great masses ; for 
the preparation of aluminium it must, however, be reduced to a 
very fine powder. 

" In my experiments on the preparatipn of aluminium, which 
were performed in company with M. Weber, and with his most 
zealous assistance, I made use of small iron crucibles, i ^ 
inches high and i^ inches upper diameter, which I had cast 
here. In these I placed the finely divided cryolite between 
thin layers of sodium, pressed it down tight, covered it with a 


good layer of potassium chloride (KCl), and closed the cruc- 
ible with a well-fitting porcelain cover. I found potassium 
chloride the most advantageous flux to employ ; it has the low- 
est specific gravity of any which could be used, an important 
point when the slight density of the metal is taken into consid- 
eration. It also increases the fusibility of the sodium fluoride. 
I usually employed equal weights of cryolite and potassium 
chloride, and for every five parts of cryolite two parts of so- 
dium. The most fitting quantity for the crucible was found to 
be ten grammes of powdered cryolite. The whole was raised 
to a strong red heat by means of a gas-air blowpipe. It was 
found most advantageous to maintain the heat for about half an 
hour, and not longer, the crucible being kept closely covered 
the whole time ; the contents were then found to be well fused. 
When quite cold the melted mass is removed from the crucible 
by means of a spatula ; this is facilitated by striking the outside 
with a hammer. The crucible may be employed several times, 
at last it is broken by the hammer blows. The melted mass is 
treated with water, when at times only a very minute evolution 
of hydrogen gas is observed, which has the same unpleasant 
odor as the gas evolved during solution of iron in hydrochloric 
acid. The carbon contained in this gas is derived from a very 
slight trace of naphtha adhering to the sodium after drying it. 
On account of the difficult solubility of sodium fluoride, the 
mass is very slowly acted on by water, although the insolubility 
is somewhat diminished by the presence of the potassium chlo- 
ride. After twelve hours the mass is softened so far that it 
may be removed from the liquid and broken down in a porce- 
lain mortar. Large globules of aluminium are then discovered, 
weighing from 0.3 to 0.4 or even 0.5 grammes, which may be 
separated out. The smaller globules cannot well be separated 
from the undecomposed cryolite and the alumina always pro- 
duced by washing, owing to their being specifically lighter than 
the latter. The whole is treated with nitric acid in the cold. 
The alumina is not dissolved thereby, but the little globules 
then first assume their true metallic lustre. They are dried and 


rubbed on fine silk muslin ; the finely powdered, undecom- 
posed cryolite and alumina pass through, while the globules 
remain on the gauze. The mass should be treated in a plati- 
num or silver vessel; a porcelain vessel would be powerfully 
acted on by the sodium fluoride. The solution, after standing 
till clear, may be evaporated to dryness in a platinum capsule, 
in order to obtain the sodium fluoride, mixed however with 
much potassium chloride. The small globules may be united 
by fusion in a small well-covered porcelain crucible, under a 
layer of potassium chloride. They cannot be united without a 
flux. They cannot be united by mere fusion, like globules of 
silver, for instance, for though they do not appear to oxidize 
on ignition in the air, yet they become coated with a scarcely 
perceptible film of oxide, which prevents their running together 
into a mass. This fusion with potassium chloride is always at- 
tended with a loss of aluminium. Buttons weighing 0.85 
gramme lost, when so treated, 0.05 gramme. The potassium 
chloride when dissolved in water left a small quantity of alu- 
mina undissolved, but the solution contained none. Another 
portion of the metal had undoubtedly decomposed the potas- 
sium chloride ; and a portion of this salt and aluminium chlo- 
ride must have been volatilized during fusion (other metals, as 
copper and silver, behave in a similar manner — Pogg. Ixviii. 
287). I therefore followed the instructions of Deville, and 
melted the globules under a stratum of aluminium-sodium chlo- 
ride in a covered porcelain crucible. The salt was melted first, 
and 'then the globules of metal added to the melted mass. 
There is no loss, or a very trifling one of a few milligrammes of 
metal, by this proceeding. When the aluminium is fused 
under potassium chloride its surface is not perfectly smooth, 
but exhibits minute concavities ; with aluminium-sodium chlo- 
ride this is not the case. The readiest method of preparing 
the double chloride for this purpose is by placing a mixture of 
alumina and carbon in a glass tube, as wide as possible, and 
inside this a tube of less diameter, open at both ends, and con- 
taining sodium chloride. If the spot where the mixture is 


placed be very strongly heated, and that where the sodium 
chloride is situated, more moderately, while a current of chlo- 
rine is passed through the tube, the vapor of aluminium chlo- 
lide is so eagerly absorbed by the sodium chloride that none 
or at most a trace is deposited in any other part of the tube. 
If the smaller tube be weighed before the operation, the 
amount absorbed is readily determined. It is not uniformly 
combined with the sodium chloride, for that part which is near- 
est to the mixture of charcoal and alumina will be found to 
have absorbed the most. 

" I have varied in many ways the process for the preparation 
of aluminium, but in the end have returned to the one just de- 
scribed. I often placed the sodium in the bottom of the cruci- 
ble, the powdered cryolite above it, and the potassium chloride 
above all. On proceeding in this manner, it was observed that 
much sodium was volatilzed, burning with a strong yellow 
flame, which never occurred when it was cut into thin slices and 
placed in alternate layers with the cryolite, in which case the 
process goes on quietly. When the crucible begins to get red 
hot, the temperature suddenly rises, owing to the commence- 
ment of the decomposition of the compound ; no lowering of the 
temperature should be allowed, but the heat should be steadily 
maintained, not longer, however, than half an hour. By pro- 
longing the process a loss would be sustained, owing to the 
action of the potassium chloride on the aluminium. Nor does 
the size of the globules increase on extending the time even to 
two hours : this efifect can only be produced by obtaining the 
highest possible temperature. If the process be stopped, how- 
ever, after five or ten minutes of very strong heat, the produc- 
tion is very small, as the metal has not had sufficient time 
to conglomerate into globules, but is in a pulverulent form and 
burns to alumina during the cooling of the crucible. No ad- 
vantage is gained by mixing the cryolite with a portion of 
chloride before placing it between the layers of sodium, neither 
did I increase the production by using aluminium sodium 
chloride to cover the mixture instead of potassium chloride. I 


repeatedly employed decrepitated sodium chloride as a flux in the 
absence of potassium chloride, without remarking any important 
difference in the amount of metal produced, although a higher 
temperature is in this case required. The operations may also 
be conducted in refractory unglazed crucibles made of stone- 
ware, and of the same dimensions, although they do not resist so 
well the action of the sodium fluoride at any high heats, but fuse 
in one or more places. The iron crucibles fuse, however, 
when exposed to a very high temperature in a charcoal fire. 
The product of metal was found to vary very much, even when 
operating exactly in the manner recommended, and with the 
same quantities of materials. I never succeeded in reducing the 
whole amount of metal contained in the cryolite (which con- 
tains 13 per cent, of aluminium). By operating on 10 
grammes of cryolite, the quantity I always employed in the 
small iron crucible, the most successful result was 0.8 grm. But 
0.5 or even 0.4 grm. may be considered favorable ; many times 
I obtained only 0.3 grm., or even less. These very different 
results depend on various causes, more particularly, however, 
on the degree of heat obtained. The greater the heat, the 
greater the amount of large globules, and the less amount of 
minutely divided metal to oxidize during the cooling of the 
crucible. I succeeded once or twice in reducing nearly the 
whole of the metal to one single button weighing 0.5 grm., at 
a very high heat in a stoneware crucible. I could not always 
obtain the same heat with the blowpipe, as it depended in some 
degree on the pressure in the gasometer in the gas-works, 
which varies at different hours of. the day. The following ex- 
periment will show how great the loss of metal may be owing 
to oxidation during the slow cooling of the crucible and its con- 
tents : In a large iron crucible were placed 35 grms. of cryolite 
in alternate layers with 14 grms. of sodium, and the whole cov- 
ered with a thick stratum of potassium chloride. The crucible, 
covered by a porcelain cover, was placed in a larger earthen 
one also covered, and the whole exposed to a good heat in a 
draft furnace for one hour, and cooled as slowly as possible. 


The product in this case was remarkably small, for O.135 grm. 
of aluminium was all that could be obtained in globules. The 
differences in the amounts reduced depend also in some degree 
on the more or less successful stratification of the sodium 
with the powdered cryolite, as much of the latter sometimes 
escapes decomposition. The greater the amount of sodium 
employed, the less likely is this to be the case ; however, owing 
to the great difference in their prices, I never employed more 
than 4 grms. of sodium to 10 grms. of cryolite. In order to 
avoid this loss by oxidation I tried another method of prepara- 
tion : Twenty grms. of cryolite were heated intensely in a gun- 
barrel in a current of hydrogen, and then the vapor of 8 grms. 
of sodium passed over it. This was effected simply by placing 
the sodium in a little iron tray in a part of a gun-barrel without 
the fire, and pushing it forward when the cryolite had attained 
a maximum temperature. The operation went on very well, 
the whole being allowed to cool in a current of hydrogen. 
After the treatment with water, in which the sodium fluoride 
dissolved very slowly, I obtained a black powder consisting for 
the most part of iron. Its solution in hydrochloric acid gave 
small evidence of aluminium. The small amounts I obtained, 
however, should not deter others from making these experi- 
ments. These are the results of first experiments, on which I 
have not been able to expend much time. Now that cryolite 
can be procured at so moderate a price, and sodium by De- 
ville's improvements will in future become so much cheaper, it 
is in the power of every chemist to engage in the preparation of 
aluminium, and I have no doubt that in a short time methods 
M'ill be found affording a much more profitable result. 

" To conclude, I am of opinion that cryolite is the best 
adapted of all the compounds of aluminium for the preparation 
of this metal. It deserves the preference over aluminium-sodium 
chloride or aluminium chloride, and it might still be employed 
with great advantage even if its price were to rise considerably. 
The attempts at preparing aluminium direct from alumina have 
as yet been unattended with success. Potassium and sodium 


appear only to reduce metallic oxides when the potash and 
soda produced are capable of forming compounds with a por- 
tion of the oxide remaining as such. Pure potash and soda, 
with whose properties we are very slightly acquainted, do not 
appear to be formed in this case. Since, however, alumina 
combines so readily with the alkalies to form aluminates, one 
would be inclined to believe that the reduction of alumina by^ 
the alkali metals should succeed. But even were it possible to^ 
obtain the metal directly from alumina, it is very probable that 
cryolite would long be preferred should it remain at a moderate 
price, for it is furnished by nature in a rare state of purity, and 
the aluminium is combined in it with sodium and fluorine only, 
which exercise no prejudicial influence on the properties of the 
metal, whereas alumina is rarely found in nature in a pure state 
and in a dense, compact condition, and to prepare it on a large 
scale, freeing it from those substances which would act injuri- 
ously on the properties of the metal, would be attended with 
great difficulty. 

" The buttons of aluminium which I have prepared are so 
malleable that they may be beaten and rolled out into the 
finest foil without cracking on the edges. They have a strong 
metallic lustre. Some small pieces, not globular, however, 
were found in the bottom of the crucible, and occasionally ad- 
hering to it, which cracked on being hammered, and were dif- 
ferent in color and lustre from the others. They were evi- 
dently not so pure as the greater number of globules, and con- 
tained iron. On sawing through a large button weighing 3.8 
grammes, it could readily be observed that the metal for abotit 
half a line from the exterior was brittle, while in the interior it 
was soft and malleable. Sometimes the interior of a globule 
contained cavities. With Deville, I have occasionally observed 
aluminium crystallized. A large button became striated and 
crystalline on cooling. Deville believes he has observed reg- 
ular octahedra, but does not state this positively. According 
to my brother's examination, the crystals do not belong to any 
of the regular forms. As I chanced on one occasion to attempt 


the fusion of a large, flattened- out button of rather impure 
aluminium, without a flux, I observed, before the heat was suffi- 
cient to fuse the mass, small globules sweating out from the 
surface. The impure metal being less fusible than pure metal, 
the latter expands in fusing and comes to the surface." 

Experiments of Percy and Dick (1855). 

After the publication of Rose's results, widespread attention 
was directed toward this field, and it was discovered that some 
six months previously Dr. Percy, in England, had accomplished 
almost similar results, and had even shown a specimen of the 
metal to the Royal Institution, but with the singular fact of ex- 
citing very little attention. These facts are stated at length in 
the following paper written by Allan Dick, Esq., which ap- 
peared in November, 1855, two months after the publication of 
H. Rose's paper: — * 

" In the last number of this magazine was the translation of a 
paper by H. Rose, of Berlin, describing a method of preparing 
aluminium from cryolite. Previously, at the suggestion of Dr. 
Percy, I had made some experiments on the same subject in 
the metallurgical laboratory of the School of Mines ; and as the 
results obtained agree very closely with those of Mr. Rose, it 
may be interesting to give a short account of them now, though 
no detailed description was published at the time, a small piece 
of metal prepared from cryolite having simply been shown at 
the weekly meeting of the Royal Institution, March 30, 1855, 
accompanied by a few words of explanation by Faraday. 

" Shortly after the publication of Mr. Deville's process for 
preparing aluminium from aluminium chloride, I tried along 
with Mr. Smith to make a specimen of the metal, but we found 
it much more difficult to do than Deville's paper had led us to 
anticipate, and had to remain contented with a much smaller 
piece of metal than we had hoped to obtain. It is, however, 
undoubtedly only a matter of time, skill, and expense to join 

* Phil. Mag., Nov. 1855. 


successful practice with the details given by Deville. Whilst 
making these experiments, Dr. Percy had often requested us to 
try whether cryolite could be used instead of the chlorides, but 
some time elapsed before we could obtain a specimen of the 
mineral. The first experiments were made in glass tubes 
sealed at one end, into which alternate layers of finely powdered 
cryolite and sodium cut into small pieces were introduced, and 
covered in some instances with a layer of cryolite, in others by 
sodium chloride. The tube was then heated over a gas blow- 
pipe for a few minutes till decomposition had taken place, and 
the product was melted. When cold, on breaking the tube, it 
was found that the mass was full of small globules of aluminium, 
but owing to the specific gravity of the metal and flux being 
nearly alike, the globules had not collected into a button 
at the bottom. To effect this, long-continued heat would be 
required, which cannot be given in glass tubes owing to the 
powerful action of the melted fluoride on them. To obviate 
this difficulty, a platinum crucible was lined with magnesia 
by ramming it in hard and subsequently cutting out all but 
a lining. In this, alternate layers of cryolite and sodium were 
placed, with a thickish layer of cryolite on top. The cruci- 
ble was covered with a tight-fitting lid, and heated to redness 
for about half an hour over a gas blowpipe. When cold it was 
placed in water, and after soaking for some time the contents 
were dug out, gently crushed in a mortar, and washed by de- 
cantation. Two or three globules of aluminium, tolerably large 
considering the size of the experiment, were obtained, along 
with a large number of very small ones. The larger ones were 
melted together under potassium chloride. Some experiments 
made in iron crucibles were not attended with the same success 
as those of Rose : no globules of any considerable size re- 
mained in the melted fluorides ; the metal seemed to alloy on 
the sides of the crucible, which acquired a color like zinc. It 
is possible that this difference may have arisen from using a 
higher temperature than Rose, as we made these experiments 
in a furnace, not over the blowpipe. Porcelain and clay cru- 


cibles were also tried, but laid aside after a few experiments, 
owing to the action of the fluorides upon them, which in most 
cases was sufificient to perforate them completely." 

Deville's Methods (1856-8). 

* " I have repeated and confirmed all the experiments of Dr. 
Percy and H. Rose, using the specimens of cryolite which I 
obtained from London through the kindness of MM. Rose and 
Hofmann. I have, furthermore, reduced cryolite mixed with 
sodium chloride by the battery, and I believe that this will be 
an excellent method of covering with aluminium all the other 
metals, copper in particular. Anyhow, its fusibility is consid- 
erably increased by mixing it with aluminium-sodium chloride. 
Cryolite is a double fluoride of aluminium and sodium, con- 
taining 13 per cent, of aluminium and having the formula 
AlFj.sNaF. I have verified these facts myself by many 

" In reducing the cryolite I placed the finely pulverized mix- 
ture of cryolite and sodium chloride in alternate layers with 
sodium in a porcelain crucible. The uppermost layer is of 
pure cryolite, covered with salt. The mixture is heated just to 
complete fusion, and after stirring with a pipe-stem, is let cool. 
On breaking the crucible, the aluminium is often found united 
in large globules, easy to separate from the mass. The metal 
always contains silicon, which increases the depth of its natural 
blue tint and hinders the whitening of metal by nitric acid, be- 
cause of the insolubility of the silicon in that acid. M. Rose's 
metal is very ferruginous. I have verified all M. Rose's ob- 
servations, and I agree with him concerning the return of 
metal, which I have always found very small. There are al- 
ways produced in these operations brilliant flames, which are 
observed in the scoria floating on the aluminium, and which 
are due to the gas burning and exhaling a very marked odor 
of phosphorus. In fact, phosphoric acid exists in cryolite, as 

* Ann. de Chim. et de Phys. [3], xlvi. 451. 


one may find by treating a solution of the mineral in sulphuric 
acid with molybdate of ammonia, according to H. Rose's re- 

" M. Rose has recommended iron vessels for this operation, 
because of the rapidity with which alkaline fluorides attack 
earthen crucibles and so introduce considerable silicon into the 
metal. Unfortunately, these iron crucibles introduce iron into 
the metal. This is an evil inherent in this method, at least in 
the present state of the industry. The inconveniences of this 
method result in part from the high temperature required to 
complete the operation, and from the crucible being in direct 
contact with the fire, by which its sides are heated hotter than 
the metal in the crucible. The metal itself, placed in the lower 
part of the fire, is hotter than the slag. This, according to my 
observations, is an essentially injurious condition. The slag 
ought to be cool, the metal still less heated, and the sides of 
the vessel where the fusion occurs ought to be as cold as pos- 
sible. The yield from cryolite, according to Rose's and my own 
observations, is also very small. M. Rose obtained from lo of 
cryolite and 4 of sodium about 0.5 of aluminium. This is due 
to the affinity of fluorine for aluminium, which must be very 
strong not only with relation to its affinity for sodium, but even 
for calcium, and this affinity appears to increase with the tem- 
perature, as was found in my laboratory. Cryolite is most 
convenient to employ as a flux to add to the mixture which is 
fused, especially when operating on a small scale. 

"The argument which decided the company at Nanterre not 
to adopt the method of manufacture exclusively from cryolite 
was the report of M. de Chancourtois, mining engineer, who 
had just returned from a voyage to Greenland. According to 
the verbal statements of this gentleman, the gite at Evigtok is 
accessible only during a very short interval of time each year, 
and, because of the ice fields, can only be reached then by a 
steamboat. The workmen sent from Europe to blast and load 
up the rock have scarcely one or two months of work possible. 
The local workmen remain almost a whole year deprived of all 


communication with the rest of the world, without fresh provi- 
sions or fuel other than that brought from Europe in the short 
interval that navigation is open. The deposit itself, which is 
scarcely above sea-level, can be easily worked with open roof, 
but the neighborhood of the sea in direct contact with the vein, 
the unorganized manner of working, and the lack of care in 
keeping separate the metalliferous portions of the ore — all com- 
bine to render the mineral very costly, and further developments 
underground almost impossible. 

" It is therefore fortunate that cryolite is not indispensable, 
for no one would wish to establish an industry based on the 
employment of a material which is of such uncertain supply." 

Tissier Bros.' Method (1857). 

The process adopted in the works at Amfreville, near Rouen, 
directed by Tissier Bros., is essentially that described by Percy 
and Rose. The method of operating is given by the Tissier 
Bros, themselves in their book as follows : 

"After having finely powdered the cryolite, it is mixed with 
a certain quantity of sodium chloride (sea salt), then placed be- 
tweeen layers of sodium used in the proportions given by M. 
Rose, in large refractory crucibles. These are heated either in 
a reverberatory furnace or in a wind furnace capable of giving 
a temperature high enough to melt the fluoride of sodium pro- 
duced by the reaction. As the sodium fluoride requires a 
pretty high temperature to fuse it, the heat will necessarily be 
higher than that required in the reduction of the double 
chloride of aluminium and sodium. When the contents of the 
crucible are melted, so as to be quite liquid, the fusion is poured 
into cast-iron pots, at the bottom of which the aluminium col- 
lects in one or several lumps." 

Tissier Bros, claimed the following advantages for the use of 
cryolite : 

" Cryolite comes to us of a purity difficult to obtain with the 
double chloride of aluminium and sodium, to which it exactly 
corresponds ; and since, thanks to the perfection we have at- 


tained in using it, the return of aluminium is exactly correspon- 
dent to the amount of sodium used in reduction, it is easily seen 
what immense advantages result from its employment. The 
double chloride deteriorates in the air, it gives rise in the works 
to vapors more or less deleterious and corrosive, and its price 
is always high. Cryolite can be imported into France at a 
price so low that we have utilized it economically for making 
commercial carbonate of soda ; it remains unaltered in the air, 
emits no deleterious vapors, and its management is much more 
easy than that of the double chloride. Moreover, on compar- 
ing the residues of the two methods of reduction, the manufac- 
ture from double chloride leaves sodium chloride, almost with- 
out value, while the manufacture from cryolite leaves sodium 
fluoride, which may be converted for almost nothing into 
caustic soda or carbonate, and so completely cancels the cost 
of the cryolite from the cost of aluminium. The most serious 
objection which can be made to using cryolite is that the 
sources of the mineral being up to the present very limited, the 
future prospect of aluminium Hes necessarily in the utilization 
of clays and their transformation into aluminium chloride ; but, 
admitting that other sources of cryolite may not be discov- 
ered hereafter, the abundance of those which exist in Greenland 
will for a long time to come give this mineral the preference in 
the manufacture of aluminium." 

The most serious difhculty which this process had to meet, 
and which it could not overcome, was the high content of sili- 
con in the metal produced. A specimen of their aluminium 
made in 1859 contained 4.4 per cent, of silicon alone (see p. 
54, Analysis 7). The firm at Rouen went out of business about 
1863 or 1865 — I am unable to give the exact date. From that 
time until quite recently, it has been considered that the best 
use of cryolite is as a flux in the preparation of aluminium from 
aluminium-sodium chloride, in which case the slag is not sodium 
fluoride, but aluminium fluoride, which acts but slightly on the 
containing vessel. 


Wohler' s Modifications (1856). 

Wohler suggested the following modifications of Deville's 
process of reducing cryolite in crucibles, by means of which 
the reduction can be performed in an earthen crucible without 
the metal produced taking up silicon. 

*"The finely pulverized cryolite is mixed with an equal 
weight of a flux containing 7 parts sodium chloride to 9 parts 
potassium chloride. This mixture is then placed in alternate 
layers with sodium in the crucible, 50 parts of the mixture to 
10 of sodium, and heated gradually just to its fusing point. 
The metal thus obtained is free from silicon, but only one-third 
of the aluminium in the cryolite is obtained." In spite of the 
small yield, this method was used for some time by Tissier 

Gerhard's Furnace (1858). 

This furnace was devised for the reduction of aluminium 
either from aluminium-sodium chloride or from cryolite, the 
object being to prevent loss of sodium by ignition. It was in- 
vented and patented by W. F. Gerhard. | " It consists of a 
reverberatory furnace having two hearths, or of two crucibles, 
or of two reverberatory furnaces, placed one above the other 
and communicating by an iron pipe. In the lower is placed a 
mixture of sodium with the aluminium compound, and in the 
upper a stratum of sodium chloride, or of a mixture of this 
salt and cryolite, or of the slag obtained in a previous ope- 
ration. This charge, when melted, is made to run into the 
lower furnace in quantity sufficient to completely cover the 
mixture contained therein, and so to protect it from the air. 
The mixture thus covered is reduced as by the usual operation." 

Whether a furnace was ever put up and operated on this 
principle, the author cannot say. It is possible that it may 
have been used in the English manufactory started in 1859 
at Battersea near London. (See p. 18.) 

* Ann. der Chem. und Pharm. 99,255. 
t Eng. Pat., 1858, No. 2247. 


Thompson and White's Patent (1887). 

* J. B. Thompson and W. White recommend heating a mix- 
ture of 3 parts sodium and 4 parts cryolite to 100°, whereby 
the sodium becomes pasty and the whole can be well kneaded 
together with an iron spatula. When cold, 4 parts of alumin- 
ium chloride are added, and the mixture put into a hopper on 
top of a well-heated reverberatory furnace, with a cup-shaped 
hearth. The charge is dropped into the furnace and the 
reaction takes place at once. To produce alloys, this patent 
claims 16 parts of cryolite are mixed with 5 parts of sodium, 
the metal added before reduction and the mixture treated as 
above, by which means explosions are avoided. The pre- 
liminary heating to 100° is effected in a jacketed cast-iron pot 
connected with a circulating boiler. 

Hampe's Experiment (1888). 

t Dr. W. Hampe failed to produce aluminium bronze by treat- 
ing cryolite with sodium in the presence of copper. A mixture of 

Finely divided copper 44 grammes, 

Sodium, in small pieces, 15 " 

Finely powdered cryolite 100 " 

was melted rapidly in a carbon-lined crucible. There were no 
sounds given out such as usually accompany other reductions 
by sodium, but much sodium vapor was given off. The copper 
button contained only traces of aluminium. 

Netto's Process (1887). 

Dr. Curt Netto, of Dresden, patented in England and Ger- 
many, in spring and autumn of 1887, processes for producing 
sodium and potassium and methods of using them in producing 
aluminium. His experiments were made in conjunction with 
Dr. Salomon, of Essen, and the fact that the experimental ap- 

* English patent 8427, June 11, 1887. 
tChemiker Zeitung, (Cothen), xii, p. 391. 


paratus was put up in Krupp's large steel works at Essen gave 
rise to reports that the latter had taken up the manufacture of 
aluminium by some new and very successful process, intending 
to use it for alloys in making cannon. * 

In the latter part of 1888 was formed the Alliance Aluminium 
Co. of London, England, capitalized at ;£'5oo,ooo, purposing to 
manufacture sodium and aluminium, and owning the English, 
French, German, and Belgian patents of Dr. Netto for the pro- 
duction of those metals, also the processes of a Mr. Cunning- 
ham for the same purpose, also a process for the production of 
artificial cryolite by the regeneration of slag (provisionally pro- 
tected by its inventor, Mr. Forster, of the Lonesome Chemical 
Works, Streatham), and, lastly, a process invented by Drs. 
Netto and Salomon by which aluminium can be raised to the 
highest standards of purity on a commercial scale. A note 
accompanying the above announcement stated that the exhaust- 
ive experiments made at Essen had satisfactorily demonstrated 
the practicability of the processes, and that the company had 
already contracted with the cryolite mines of Greenland for all 
the cryolite they would need. 

In June, i888,t the Alliance Aluminium Company had in 
operation a small aluminium plant at King's Head Yard, Lon- 
don, E. C, and when the process was in continuous opera- 
tion the cost of the metal was set down at 6 shillings per 
pound. It is probable that the metal exhibited in the Paris 
Exposition of 1889 was produced at this place. 

In April, 1889,! ten acres of ground had been leased at 
Hepburn on which to produce sodium by Capt. Cunningham's 
process. The sodium produced was to be sent to Wallsend to 
be used by the Alliance Aluminium Company, who were erect- 
ing a large works at that place. 

As for Capt. Cunningham's sodium processes, they are ap- 
parently identical with Dr. Netto's. Cunningham's aluminium 

♦American Register, Paris, August, 1888. 
t Engineering, June i, 1888. 
X E. and M. J., April 27, 1889. 



process* consists in melting the sodium to be used with lead, 
in order to facihtate the submerging of the sodium under the 
molten aluminium salt. The alloy is cast into bars and added 
piece by piece to the bath of molten aluminium salt on the 
hearth of a reverberatory furnace. After the reaction the mix- 
ture separates by specific gravity into lead, containing a little 
aluminium, and aluminium containing a little lead, the slag 
floating on top of all. Aluminium is known to have so small an 
attraction for lead that this result becomes possible. 

Fig. 26. 

Dr. Netto recommends several processes, the one used at 
London being the following: — | 

One hundred parts of cryolite and 30 to 100 parts of sintered 
sodium chloride are melted at a red heat in a well-covered clay 
crucible. (Another arrangement, and apparently a better, is to 
melt this mixture on the hearth of a reverberatory furnace and 
tap it into a deep, conical ladle, in which the succeeding opera- 
tions proceed as about to be described. See Fig. 26.) As soon 
as the bath is well fused, 35 parts of sodium at the end of a rod, and 

* English Patent, 16727, Dec. 5, 1887. 

t German Patent (D. R. P.), 45198, March 26, 1887. 


covered over by a perforated concave plate is pushed quickly 
down to the bottom of the crucible. The plate mentioned fits 
across the whole section of the crucible at its lower part, so that 
the fusible, easily volatile sodium, being vaporized, is divided into 
very fine streams as it passes upwards through the bath, and is 
all utilized before it reaches the surface. In this way the re- 
action is almost instantaneous, and the contents can be poured 
out at once into iron pots, where, on cooling, the metal is found 
as a large lump at the bottom. 

It is further observed that to avoid explosions on introducing 
the sodium it should have in it no cavities which might contain 
moisture or hydrocarbons. In consequence of the reaction be- 
ing over so quickly, and the heat set free in the reduction, the 
syrupy fusion becomes thin as water, and the aluminium dis- 
seminated through the mass collects together completely, so 
that the slag contains no particles visible to the eye. Since the 
reduction, pouring, and cooling take place so quickly, the alu- 
minium is not noticeably redissolved by the bath, thus insuring 
a high return of metal. By using 35 parts of sodium to 100 
parts of cryolite, 10 parts of aluminium are obtained. Since 
the cryolite contains 13 per cent, of aluminium, the return is 
yy per cent, of the amount of metal in the cryolite; since 35 
parts of sodium should theoretically displace 14 parts of alu- 
minium, the return is 71 per cent, of the amount which the 
sodium should produce. Dr. Netto claims that this is double 
the return formerly obtained from cryolite. The metal pro- 
duced is said to be from 98.5 to 99 per cent. pure. 

The apparatus erected at Krupp's works at Essen, which was 
described by the newspapers as similar to a Bessemer converter, 
was constructed and operated as follows : A large iron cylinder 
is pivoted at the centre in a manner similar to a Bessemer con- 
verter. Passing through the centre of the cylinder, longitudi- 
nally, is a large iron tube in which generator gas is burnt to 
heat the vessel.. To heat it up, it is placed erect, connection 
made with the gas-main, while a hood above connects with the 
chimney. On top of the cylinder, a close valve communicates 


with the interior, for charging, and at the other end is a tap- 
hole. The charge of cryoHte being put in, the flame is passed 
through the central tube until the mineral is well fused. Then 
soHd or melted sodium is passed in at the top, the valve is 
screwed tight, the gas shut off, and the whole cylinder is rotated 
several times until reduction is complete, when it is brought 
upright, the tap-hole opened and slag and metal tapped into a 
deep iron pot, where they separate and cool. Aluminium thus 
made could not but contain much iron, even up to 14 per cent., 
it is said, which would prevent its use for any purpose ex- 
cept alloying with iron. To procure pure aluminium, the ves- 
sel would have to be properly fettled. 

Dr. Netto also devised an arrangement similar to Heaton's 
apparatus for making steel. It consisted of a large, well-lined 
vessel on trunnions, the bottom of which was filled to a certain 
depth with sodium, then a perforated aluminium plate placed 
like a false bottom over it. On pouring molten cryolite into 
the vessel, the aluminium plate prevented the sodium from rising 
en masse to the surface of the cryolite. After the reaction was 
over, the vessel was tilted and the slag and metal poured out 
into iron pots. 

The modification of the crucible method appears to have 
been the most feasible of Netto's processes, and was probably 
that used at Wallsend by the Alliance Aluminium Com- 
pany. Outside estimates of the cost of aluminium to this com- 
pany placed it at $1.50 to $2 per pound. They were selling in 
the latter part of 1889 at 11, 13, and 15 shillings per pound, 
according to quality, but were forced out of the business by the 
electrolytic processes in 189 1. 

Grabau's Process (1887). 

There is only one patentee claiming particularly the reduc- 
tion of aluminium fluoride by sodium — Ludwig Grabau, of 
Hannover, Germany. His patents on this subject are im- 
mediately preceded by others on a method of producing the 
aluminium fluoride cheaply, which are described on p. 171, and 


the inventor has patented a process which will furnish him with 
cheap sodium. Mr. Alexander Siemens is authority for the 
statement that a plant was in operation in the spring of 1889, 
in Hannover, producing aluminium by this process on a com- 
mercial scale. The principal object of Mr. Grabau's endeavors 
has been to produce metal of a very high degree of purity. To 
this end every precaution is taken to procure pure materials 
and to prevent contamination during reduction. We will quote 
from a paper written by Mr. Grabau* and also from his speci- 
fications, f the following explanation of the process : 

" The purifying of impure aluminium is accompanied by so 
many difficulties that it appears almost impossible. It is there- 
fore of the greatest importance to so conduct the operation that 
every impurity is excluded from the start. Molten aluminium 
compounds, whether a flux is added or not, attack any kind of 
refractory vessels and become siliceous if these vessels are made 
of chamotte or like materials, or if made of iron they become 
ferruginous. These impurities are reduced in the further pro- 
cesses and pass immediately into the aluminium as iron, silicon, 
etc. Evidently the case is altered if an aluminium compound 
which is infusible can be used advantageously. Aluminium 
fluoride is infusible and also retains its pulverized condition 
when heated up to the temperature needed for its use ; it can 
therefore be heated in a vessel of any kind of refractory mater- 
ial, or even in a metallic retort, without danger of taking up any 

" Further, it is necessary for succeeding in producing alu- 
minium that the reduced metal thall unite to a large body after 
the reduction. For this purpose all previous processes use 
fluxes, and usually cryolite. But cryolite is impure, and there- 
fore here is a source of many of the impurities in commercial 
aluminium. Dr. K. Kraut, of Hannover, has observed that, 
according to the recent analyses of Fresenius and Hintz, com- 

* Zeitschrift fur angewandte, Chemie, 1889, vol. 6. 

t German Pat. (D. R. P.) 47°3'. Nov. 15, 1887. English Pat. 15593, Nov. 14, 
1887. U. S. Patents 386704, July 24, l888; and 400449, April 2, 1889. 


mercial cryolite contains 0.80 to 1.39 per cent, of silicon, and 
o.ii to 0.88 per cent, of iron, and that these impurities inter- 
penetrate the mineral in such a manner as to be often only vis- 
ible under the microscope, and therefore totally impossible of 
removal by mechanical means. It is thus seen that the avoid- 
ance of the use of any Rux is of great importance as far as 
producing pure metal is concerned, as well as from an eco- 
nomic standpoint. 

" By the following process it is also possible to reduce alu- 
minium fluoride by ?odium without the vessel in which reduc- 
tion takes place being attacked either by the aluminium-sodium 
fluoride formed or by the reduced aluminium. For this pur- 
pose the aluminium fluoride and sodium are brought together 
in such proportions that after the reaction there is still suffi- 
cient aluminium fluoride present to form with the sodium flu- 
oride resulting from the reaction a compound having the com- 
position of cryolite. The reaction, therefore, will be 


" Using these proportions, the aluminium fluoride must be 
previously warmed up to about 600° C, in order that when it is 
showered down upon the melted sodium the reaction may com- 
mence without further application of heat. The aluminium 
fluoride remains granular at this temperature and therefore re- 
mains on top of the melted sodium, like saw-dust or meal upon 
water, and under its protection the reaction proceeds from 
below upwards — an important advantage over the usual method 
of pouring molten aluminium compounds on to sodium, in 
which the lighter sodium floats to the top and burns to waste. 
If solid sodium is used in my process the aluminium fluoride 
must be somewhat hotter on being poured into the reduction 
vessel, or about 700° C. For carrying out the process the re- 
duction vessel must be artificially cooled, so as to form a lining 
by chilling some of the aluminium-sodium fluoride formed by 
the reaction, on the inner walls. This lining is in no wise 
further attacked by the contents of the vessel, nor can it evi- 
dently supply to them any impurity. 



Fig. 27. 


"The furnace A (Fig. 27) with grate B and chimney C, 
serves for heating the iron retorts D and E, which are coated 
with chamotte and protected from the direct action of the flame 
by brick work. The vessel D serves for heating the aluminium 
fluoride, and is provided with a damper or sliding valve be- 
neath. The sodium is melted in E, and can be emptied out by 
turning the cock k. The water-jacketed reduction vessel is 
mounted on trunnions to facilitate emptying it. The retorts 
are first heated dark red-hot, and D is filled with the conven- 
ient quantity of aluminium fluoride. When this has become 
red hot, as is shown by a small quantity of white vapor is- 
suing from it, the required quantity of sodium is put into E. 
This melts very quickly, and is then immediately run into the 
reduction vessel by opening the stop-cock k. As soon as it is 
transferred, the slide at the base of the retort D is pulled out 
and the whole quantitity of aluminium fluoride falls at once 
upon the sodium, and the reaction begins. As before re- 
marked, the granular form of the aluminium fluoride keeps it 
on top of the sodium, so that the latter is completely covered 
during the whole reaction. This prevents almost altogether 
any waste of sodium by volatilization. Dr. K. Kraut testifies 
to an operation which he witnessed in which the return showed 
83 per cent, of the sodium to have been utilized. An efficiency 
in this respect of over 90 per cent, has been occasionally 
reached, while the average is 80 to 90. Ad. Wurtz states that 
the average for several years' working of the Deville process 
showed only 74.3 per cent, of the quantity of aluminium pro- 
duced which the sodium used could have given. 

"During the reaction a very high temperature is developed, 
so that the cryolite formed becomes very fluid, but is chilled 
against the sides of the vessel to a thickness of a centimetre or 
more. This crust is a poor conductor of heat, and is neither 
attacked by the fluid cryolite nor by the aluminium. In con- 
sequence of the great fluidity of the bath, it is possible for the 
aluminium to unite into a body without the use of any flux, 
The reaction being over, which is accomplished with the above 


proportions of materials in a few seconds, and the vessel hav- 
ing been shaken briskly backwards and forwards a few times to 
facilitate the settling of the aluminium, the whole is turned on 
the trunnions and emptied into a water-jacketed iron pot, where 
it cools. The crust of cryolite inside the reduction vessel is 
left there, and the apparatus is ready for another operation." 

M. Grabau, in a private communication to the author, sums 
up the advantages of his process, including the production of 
the aluminium fluoride, as follows : — 

1. The process is not dependent on natural cryolite, which is 
expensive, impure and not easily purified. 

2. The raw material — aluminium sulphate — can be procured 
in large quantities and of perfect purity. 

3. The aluminium fluoride is produced by a wet process, 
which offers no difficulties to production on a large scale. 

4. The fluorspar may be completely freed from foreign 
metals by washing with dilute acid ; any silica present is not 
injurious, as it remains undissolved in the residue during the 

5. The cryolite formed in each reduction contains no impur- 
ities, and an excess of it is produced which can be sold. 

6. The reduction of aluminium fluoride by my method gives 
a utilization of 80 to 90 per cent, of the sodium used, which is 
much more than can be obtained by other processes. 

7. Aluminium fluoride is infusible, and can therefore be 
heated in a vessel of any refractory material without taking up 
any impurities. It is also unchanged in the air, and can be 
kept unsealed for any length of time without deteriorating in 
the least. 

8. No flux has to be added for reduction, the use of impure 
flux being a frequent cause of impurity of the metal. 

In point of fact, M. Grabau has succeeded in producing sev- 
eral hundred pounds of aluminium averaging over 99^ per 
cent. pure. Dr. Kraut reports an analysis of an average speci- 
men with 99.62 per cent, of aluminium (see Analysis 20, p. 54), 
and metal has been made as pure as 99.8 per cent., a piece of 


which has been kindly forwarded the author by M. Grabau, and 
I freely admit it to be as fine a specimen of aluminium as I have 
ever seen. Metal of such purity and fineness can only be made 
by the electrolytic processes at considerable expense over ordi- 
nary best commercial metal, since they must use chemically pure 
alumina and extra precautions as to the purity of the electrodes. 
It therefore appears possible that M. Grabau may yet be able 
to manufacture and sell this extra quality aluminium, if he can 
bring its price anywhere near that of the electrolytic aluminium. 
The Grabau Aluminium Werke, Trotha, Germany, operates the 
Grabau processes, and does a business in manufacturing and 
selling sodium. 



As preliminary to the presentation of the various electrolytic 
methods which have been proposed or , used, it may be profit- 
able to review briefly the principles of electro-metallurgy as 
they apply to the decomposition of aluminium compounds. 
Some additional observations have been already included in 
Chapter VIII. 

The atomic weight of aluminium being 27, its chemical 
equivalent, or the weight of it equal in consbining power to one 
part of hydrogen, is 9. Therefore a current of quantity sufiS- 
cient to liberate i part of hydrogen in a certain time would 
produce 9 parts of aluminium in the same time, according to 
the fundamental law of electric decomposition. It has been 
determined that a current of i ampere acting for one second, 
liberates 0.00001035 grammes of hydrogen; therefore it will 
produce or set free from combination in the same time, 
0.00009315 grammes of aluminium. This is the electro-chemical 
equivalent of aluminium. Now, from thermo-chemical data we 
know that the amount of energy required to set free a certain 
weight of aluminium will vary with the compound from which 
it is produced ; but the above equivalent is independent of the 
compound decomposed, therefore there must be some varying 
factor connected with the quantity of the current to account 
for the different amounts of work which the current does in 
decomposing diflferent compounds of the same element. This 
is exactly in accordance with the principles of the mechanical 
or thermal equivalent of the electric current, for the statement 
" a current of one ampere," while it expresses a definite quan- 



tity of electricity, yet carries no idea of the energy represented 
by that current; we must know against what resistance or with 
what electro- motive force that quantity is moved, and then we 
can calculate its mechanical equivalent. Now, a current of i 
ampere flowing against a resistance of i ohm, or in other 
words, with a moving force or intensity of i volt, represents a 
quantity of energy in one second equal to 0.00024 calories of 
heat or to O.I kilogrammetre of work, and is therefore nearly 
^ijf of a horse power. Therefore we can calculate the theoreti- 
cal intensity of current necessary to overcome the affinities of 
any aluminium compound for which we know the appropriate 
thermal data. For instance, when aluminium forms its chlo- 
ride (see p. 228) ^ — — — =5,960 calories are developed per 

kilo of aluminium combining ; consequently the liberating of 
0.00009315 grammes of aluminium (its electro-chemical equiva- 
lent), requires the expenditure of an amount of energy equal to 
0.00009315 X 5.960 = 0.000555 calories. Since a current of 
I ampere at an intensity of i volt represents only 0.00024 cal- 
ories, the intensity of current necessary to decompose alumin- 
ium chloride is theoretically — '- — ^= 2.3 volts. In a simi- 


lar manner we can calculate that to decompose alumina would 

, . ^. , , 391600 0.00000009315 

require an electro-motive force of x Z^-^ = 

54 0.00024 

2.8 volts. These data would apply only to the substances 

named, in a fused anhydrous state ; with hydrated aluminium 

chloride in solution, a far. greater electro-motive force would be 

necessary. If we had the thermal data we could also calculate 

the intensity of current necessary to decompose the sulphate, 

nitrate, acetate, etc., in aqueous solution; but failing these, we 

can reason from analogy that it would be several volts in each 


It must be noted also, that these heats of combination are 

those observed at ordinary temperatures. To find the voltage 

required at high temperatures we must make allowance for the 


weakening of the affinities, that is, for the decrease in heat of 
combination as the temperature rises. When the rate of this 
decrease is known, the true voltage required at any temperature 
can be calculated. Often the operation is reversed ; that is, the 
minimum electro-motive force required to effect decomposition 
is accurately measured at different temperatures, and from 
these data the variation in the heat of formation with the tem- 
perature can be calculated by reversing the above calculations. 
For example of this see p. 239. 

To utilize such calculations, we must bear in mind exactly 
what they represent. To decompose fused aluminium chloride, 
for instance, not only must the current possess an intensity of 
2.3 volts, but it must in addition have power enough above this 
to overcome the transfer resistance of the electrolyte ; i. e., to 
force the current through the bath from one pole to the other. 
So, then, 2.3 volts would be the absolute minimum of intensity 
which would produce decomposition, and the actual intensity 
practically required would be greater than this, varying with 
the distance of the poles apart and the temperature of the 
bath as far as it affects the conducting power of the electrolyte. 
From this it would immediately follow that if the substance to 
be decomposed is an absolute non-conductor of electricity, no 
intensity of current will be able to decompose it. If, on the 
other hand, the substance is a conductor and the poles are 
within reasonable distance, a current of a certain intensity will 
always produce decomposition. The objection is immediately 
made that in most cases no metal is obtained at all, which is 
true not because none is produced, but because it is often dis- 
solved by secondary actions as quickly as it is produced. I 
need but refer to the historic explanation of the decomposition 
of caustic soda in aqueous solution, although we have cases 
hardly parallel to this in which the electrolyte itself dissolves 
the separated metal. 

How about the case of aqueous solutions ? Water requires a 
minimum electro-motive force of 1.5 volts to decompose it, 
and hence a prominent electrician remarked of a compound 


which theoretically required over 2 volts that its decomposition 
in aqueous solution would involve the decomposition of the 
water and therefore was impossible. This remark is only partly 
true ; for caustic soda requres over 2 volts, yet if mercury is 
present to absorb the sodium as it is set free and protect it from 
the water, we will obtain sodium, while the water is decomposed 
at the same time. The truth seems to be that if two substances 
are present which require different electro-motive forces to de- 
compose them, a current of a certain intensity will decompose 
the one requiring the least force, without affecting the other at 
all; but, if it is of an intensity sufficient to decompose the 
higher compound, then the current will be divided in some ratio 
between the two, decomposing them both. This theory would 
render theoretically possible the decomposition of aluminium 
salts in aqueous solution, with a waste of power proportional to 
the amount of water decomposed at the same time ; but 
whether any aluminium would be obtained would be contingent 
on the secondary action of the water on the aluminium. Pure 
aluminium in mass is not acted upon by water, but the foil is 
rapidly eaten away by boihng water. The state of division of 
the metal, then, determines the action of water on it, and it is 
altogether probable that the reason why aluminium is not easily 
deposited from aqueous solution is that, like sodium, it is 
attacked as soon as isolated, the acidity of the solution convert- 
ing the hydrate formed back into the salt, or else simply the 
hydrate remaining. Unfortunately, mercury does not exercise 
the same function with aluminium as with sodium, for water at- 
tacks its amalgam with aluminium, and so destroys the metal. 
It is possible that if some analogous solvent could be found which 
protected the aluminium from the action of water, the deposi- 
tion from aqueous solution could be performed quite easily. 

The electro-deposition of a metal using a soluble anode is 
entirely a different affair. In this case, while the salt in solu- 
tion is decomposed, yet it is immediately regenerated by the 
acid radical set free dissolving from the anode just the same 
quantity of metal as was deposited at the cathode. This action 


exactly counterbalances the electromotive force required for 
decomposition, and leaves only the conduction resistance of the 
bath to be overcome. If this resistance is kept down so that 
the tension required to operate the bath is less than or not 
much greater than 1.5 volts, and the bath is well supplied with 
salt, there will be no water decomposed and a good plating 
may be obtained. This has been already practised on a large 
scale with aluminium. 

With anodes, however, which do not dissolve, the decom- 
position of the salt necessarily involves that of the water, if over 
1.5 volts are required to decompose it. The result of this is 
that hydrogen gas is set free liberally at the cathode along with 
the metal, and prevents the latter from depositing as a smooth, 
dense metal, causing, in fact, the production of "sponge." 
This finely-divided spongy metal is in the very best condition 
to decompose water, and the result is its oxidation. However, 
the metal can only decompose water at a certain rate ; the oxi- 
dation does not take place instantaneously. If then we de- 
posit the metal faster than it can waste away by oxidation, by 
using a current of large quantity, we may succeed in gaining 
metal and thus secure a deposit. This way of overcoming the 
difficulty is assisted by using a very strong solution, which al- 
lows a current of large quantity to pass with less resistance, 
and also leaves less water uncombined to act on the metal. 
By using these devices, Bunsen was enabled to obtain even 
metallic calcium from a solution of its chloride. 

By the density of the current we mean the number of am- 
peres passing through each unit of surface of cathode or de- 
positing surface. If the density is very great, it is necessary 
that a brisk circulation be kept up in the solution to bring fresh 
quantities of salt into the sphere of action and to remove the 
impoverished solution ; otherwise, if a deficiency of salt occurs, 
the solvent is attacked. Rapid circulation therefore favors 
regular working. 

In any electrolytic bath, the tension and density of the cur- 
rent determine which of the several ingredients present will be 


decomposed. Let us assume two compounds present, A and B, 
of which ^ is a weaker compound than By either one of them 
may be water, or they may both be anhydrous salts. In such 
a liquid bath, a current just strong enough to decompose A 
will not afifect B. When the current is just strong enough to 
decompose B, the product will be the elements of A with a 
trace of those of B/ but since the base in B is supposed to be a 
stronger one than that in A, there will be a secondary action, 
and the base of B will displace the base of A in the still unde- 
composed compound, and B will be re-formed. This will be 
the direct effect of increasing the tension of the current. How- 
ever, if the density of the current is increased at the same time, 
another principle comes into play. The base of B may be de- 
posited so quickly that it has not time to react on A, and in 
that case the deposit will contain the base of B. Also, if there 
is a much larger proportion of B than of A present, a larger 
proportion of its base will be deposited than before and escape 
re-solution. Also, other things being equal, if the compound 
B is more fluid or rather diffuses quicker in the bath than A, 
more of it will get to the electrodes and a larger proportion be 

We therefore see that when two or more compounds are 
present, the relative amounts of each to be decomposed depend 

1. Primarily, on the electro-motive force of the current, 
which may be so regulated as to decompose them one after the " 
other in the order of their electromotive force of decomposition. 

2. On the density of the current at the electrodes. That is, 
provided the tension of the current is sufficient to decompose 
the stronger compound, the greater the density of the current 
the greater the proportion of the stronger compound which 
will be decomposed, other conditions being equal. 

3. On the relative amounts of each present. Other condi- 
tions being equal, the relative amount of a compound decom- 
posed will increase as its proportion in the bath is increased. 

4. On the relative rates of diffusion, or the rapidity with 
which elements of the different compounds move towards the 


The consideration of the electrolytic processes falls naturally 
under two heads : — 

I. Deposition from aqueous solution. 
II. Non-aqueous electric processes. 


Deposition of Aluminium from Aqueovs Solution. 

The status of this question is one of the curiosities of electro- 
metallurgic science. Evidently attracted by the great reward 
to be earned by success, many experimenters have labored in 
this field, have recommended all sorts of processes, and pat- 
ented all kinds of methods. We have inventors afiSrming in 
the strongest manner the successful working of their methods, 
while other experimenters have followed these recipes, and 
tried almost every conceivable arrangement, yet report negative 
results. To show that it is quite possible that many strong af- 
firmations may be made in good faith, I have only to mention 
the fact that in March, 1863, Mr. George Gore described in the 
Philosophical Magazine some experiments by which he de- 
posited coatings of aluminium from aqueous solutions, and 
afterwards, in his text book of Electro-metallurgy, asserts that 
he knows of no successful method of doing this thing. We 
infer that Mr. Gore found that he was in error the first time. 
So, if we take the position that aluminium cannot by any 
methods so far advanced be deposited from aqueous solution 
without using a soluble aluminium anode, we will have to ad- 
mit that the proposers of many of the following processes are 
probably misled by their enthusiasm in affirming so strongly 
that they can do this thing. Yet the problem is not impossible 
of solution. 

Messrs. Thomas and Tilly* coat metals with aluminium and 
its alloys by using a galvanic current and a solution of freshly 
precipitated alumina dissolved in boihng water containing po- 
tassium cyanide, or a solution of freshly calcined alum in 

* English Patent, 1855, No. 2756. 


aqueous potassium cyanide; also from several other liquids. 
Their patent covers the deposition of the alloys of aluminium 
with silver, tin, copper, iron, silver and copper, silver, and tin, 
etc. etc., the positive electrode being of this metal or alloy. 

M. Corbelli, of Florence,* deposits aluminium by electroylz- 
ing a mixture of rock alum or sulphate of alumina (2 parts) 
with calcium chloride or sodium chloride (i part) in aqueous 
solution (7 parts), the anode being mercury placed at the 
bottom of the solution and connected to the battery by an iron 
wire coated with insulating material and dipping its uncovered 
end into the mercury. The zinc cathode is immersed in the 
solution. Aluminium is deposited on the zinc, as a blackish 
powder or as a thin, compact sheet, and the chlorine which is 
liberated at the anode unites with the mercury, forming calomel. 

J. B. Thompsonf reports that he has for over two years been 
depositing aluminium on iron, steel, and other metals, and 
driving it into their surfaces at a heat of 500° F., and also de- 
positing aluminium bronze of various tints, but declines to state 
his process. 

George Gore, Jthe noted electrician, recommended the fol- 
lowing procedure for depositing aluminium on copper, brass, or 
German silver: — 

" Take equal measures of sulphuric acid and water, or one 
part sulphuric acid, one part hydrochloric acid and two parts 
of water, put into it half an ounce of pipe clay to the pint of 
dilute acid, and boil for an hour. Take the clear, hot liquid 
and immerse in it an earthen porous cell containing sulphuric 
acid diluted with ten times its bulk of water, together with a 
rod or plate of amalgamated zinc. Connect the zinc with the 
positive wire of a Smee battery of three or four elements con- 
nected for intensity. The article to be coated, well cleaned, is 
connected with the negative pole and immersed in the hot clay 
solution. In a few minutes a fine, white deposit of aluminium 

* English Patent, 1858, No. 507. 
fChem. News, xxiv. 194 (1871). 
t Philosophical Magazine, March, 1863. 


will appear all over its surface. It may then be taken out, 
washed quickly in clean water, wiped dry, and polished. If a 
thicker coating is required, it must be taken out as soon as the 
deposit becomes dull, washed, dried, polished, and re-immersed, 
and this must be repeated at intervals as often as it becomes 
dull, until the required thickness is obtained. It is necessary 
to have the acid well saturated by boiling, or no deposit will 
be obtained." 

Mierzinski asserts that Dr. Gore was mistaken when he sup- 
posed this deposit to be aluminium, and in Gore's Text Book 
of Electro-metallurgy no mention is made of these experi- 
ments, the author thereby acknowledging the error. As to 
what the deposit could have been, we are left to conjecture, 
since no explanation has been advanced by Dr. Gore ; it may 
possibly have been silicon, mercury, or zinc, as all three of 
these were present besides aluminium. 

J. A. Jeanfon* has patented a process for depositing alumin- 
ium from an aqueous solution of a double salt of aluminium 
and potassium of specific gravity 1.161 ; or from any solution 
of an aluminium salt, such as sulphate, nitrate, cyanide, etc., 
concentrated to 20° B. at 50° F. He uses a battery of four 
pairs of Smee's or three Bunsen's cells, with elements arranged 
for intensity, and electrolyzes the solutions at 140° F. The 
first solution will decompose without an aluminium anode, but 
the others require such an anode on the negative pole. The 
solution must be acidulated slightly with acid corresponding to 
the salt used, the temperature being kept at 140° F. constantly. 

More recently!, Jeangon has obtained a patent for the fol- 
lowing process : " Subjecting a supersaturated acid solution of 
a salt of aluminium to the action of a current passed from an 
anode of aluminium in a state of division or porosity, pre- 
senting a relatively large exposure of surface within a given 
field of electric force, and a suitable metallic cathode." Further 

* Annual Record of Science and Industry, 1875. 

tU. S. Patent, 436,895; Sept. 23, i8go. (Specimens.) 


details state that the anode is to be prepared by melting alu- 
minium iri a crucible and stirring in about 30 per cent, of car- 
bon or a similar substance, and then moulding into shape. The 
electrolyte is to be maintained at a high temperature, preferably 
180° to 200° F., under which conditions a dense deposit is 

The writer has already stated that aluminium is being suc- 
cessfully electro-plated from solution using aluminium anodes; 
and, if the above process is the one which has been used in 
Philadelphia by the Harvey Filley Plating Co., then I can cer- 
tify to it having been successfully operated. The company 
mentioned refuse to say anything about the process they use, 
but I have been informed by outside parties that the above is 
the one. It is certain, however, that very good specimens of 
aluminium-plated ware have been made and sold by this firm, 
several having been examined and tested by the writer person- 
ally. A thin deposit on sheet-iron, which has been in my pos- 
session over two years, shows no sign of rust or deterioration, 
and suggests the practicability of thus making a substitute for 
tin-plate of greater durability and which will not rust. 

M. A. Bertrand * states that he deposited aluminium on a 
plate of copper from a solution of double chloride of aluminium 
and ammonia, by using a strong current, and the deposit was 
capable of receiving a brilliant polish. 

Jas. S. Haurd,t of Springfield, Mass., patented the elec- 
trolysis of an aqueous solution formed by dissolving cryolite in 
a solution of magnesium and manganous chlorides. 

John Braun { decomposes a solution of alum, of specific 
gravity 1.03 to 1.07, at the usual temperature, using an insolu- 
ble anode. In the course of the operation, the sulphuric acid 
set free is neutralized by the continual addition of alkali ; and, 
afterwards, to avoid the precipitation of alumina, a non-volatile 
organic acid, such as tartaric, is added to the solution. The 

* Chem. News, xxiv. 227. 

t U. S. Patent, 228,900, June 15, 1880. 

X German Patent, No. 28,760 (1883'!. 


intensity of the current is to be so regulated that for a bath of 
lO to 20 Htres two Bunsen elements (about 20 centimetres 
high) are used. 

Dr. Fred. Fischer* stated that Braun's proposition was con- 
trary to his experience. By passing a current of 8 to 9 volts 
and 50 amperes, using from o.i to 10 amperes per sq. centi- 
metre of cathode, with various neutral and basic aluminium 
sulphate solutions, with and without organic acids, he obtained 
no aluminium. He obtained a black deposit of copper sul- 
phide on the copper anode, which had apparently been mis- 
taken by Braun for aluminium. 

Moses G. Farmer | has patented an apparatus for obtaining 
aluminium electrically, consisting of a series of conducting cells 
in the form of ladles, each ladle having a handle of conducting 
material extending upwards above the bowl of the next suc- 
ceeding ladle ; each ladle can be heated separately from the 
rest ; the anodes are hung in the ladles, being suspended from 
the handles of the preceding ladles, the ladles themselves being 
the cathodes. 

M. L. Senet X electrolyzes a saturated solution of aluminium 
sulphate, separated by a porous septum from a solution of 
sodium chloride. A current is used of 6 to 7 volts and 4 
amperes. The double chloride, AlCl'.NaCl, is formed, then 
decomposed, and the aluminium liberated deposited on the 
negative electrode. It has later been remarked of this process 
that it has not had the wished-for success on a large scale. 

Col. Frismuth, of Philadelphia, purported to plate an alloy of 
nickel and aluminium. He used an ammoniacal solution, prob- 
ably of their sulphates. The plating certainly resembled 
nickel, but that it contained aluminium the writer is not pre- 
pared to assert. 

Baron Overbeck and H. Neiwerth, of Hannover, § have 

* Zeitschrift des Vereins Deutsche Ingenieurs, 1884, p. 557. 
t U. S. Patent, No. 315,266, April, 1885. 
t Cosmos les Mondes, Aug. 10, 1885. 
§ English Pat., Dec. 15, 1883, No. 5756: 


patented the following process : An aqueous or other solu- 
tion of an organic salt of aluminium is used, or a mixture of 
solutions which by double decomposition will yield such salt. 
Or a mixture of a metallic chloride and aluminium sulphate 
may be used, this yielding nascent aluminium chloride, which 
the current splits up immediately into aluminium and chlorine. 

Herman Reinbold* gives the following recipe, stating that it 
furnishes excellent results : 50 parts of potash alum are dis- 
solved in 300 parts of water, and to this are added 10 parts of 
aluminium chloride. The whole is then heated to 200° F., 
cooled, and then 39 parts of potassium cyanide added. A 
weak current should be used. It is stated that the plating, 
when polished, will be found equal to the best silver plating. 
" Iron," noticing this process, remarks, " there are a number of 
formulae for electro-plating with aluminium, but few appear to 
have attained to practical utility in the arts, for the reason that 
there is no special demand for such processes. All the quali- 
ties that are possessed by an electro-deposit of aluminium are 
possessed to an equal or superior degree by other metals, silver, 
nickel, platinum, etc. Furthermore, it obstinately refuses to 
take and to retain a high lustre." This criticism is a little 
overdrawn, since the one quality in which aluminium is super- 
ior to silver — not blackening by contact with sulphurous vapor 
— is not mentioned. 

Under the name of Count R. de Montegelas, of Philadelphia, 
several patents have been taken out in England for the elec- 
trolysis of aqueous solutions, which may be summarized as 
follows : — 

f Alumina is treated with hydrochloric acid, and aluminium 
chloride obtained in solution. The liquid is then placed in a 
vessel into which dip a suitable anode and a cathode of brass or 
copper. On passing an electric current through the bath the 
iron present in the liquid is first deposited, and as soon as this 
deposition ceases (as is apparent by the change of color of the 

* Jeweler's Journal, September, 1887. 

t English Patent, Aug. 18, 1886, No. 10607. 


deposit) the liquid is decanted into another similar bath, and to 
it is added about fifty per cent, by weight of the oxide of either 
lead, tin or zinc. On sending a current through this bath, 
aluminium together with the metal of the added oxide is said to 
be deposited on the cathode. 

* A rectangular vessel is divided into two unequal compart- 
ments by a vertical porous partition, into the smaller of which 
is placed a saturated solution of common salt, in which is im- 
mersed a brass or copper electrode ; into the larger is put a so- 
lution of aluminium chloride, immersed in which is an alumin- 
ium electrode. On passing the current the latter solution, 
which is normally yellow, is gradually decolorized and con- 
verted into a solution of aluminium-sodium chloride. When 
colorless, this solution is taken out and the aluminium deposited 
in a similarly arranged vessel. The double chloride solution is 
placed in the larger compartment, with an electrode of brass, 
copper, or a thin plate of aluminium, while the smaller com- 
partment contains a carbon electrode dipping into a solution of 
salt and surrounded by fragments of a mixture of salt and 
double chloride, fused together in equal parts. 

The author has been given several ounces of a very fine me- 
tallic powder said to have been made by these processes, and 
which is certainly aluminium. As I am not satisfied, however, 
that the specimen is really authentic, I feel justified in suspend- 
ing a final expression of opinion on the process. 

A Walker, of Tarnowitz, has patented the following methods 
of procedure : j 

a. Pure commercial hydrate is dissolved in nitric acid free 
from chlorine, in slight excess, and tartaric acid added. The 
Hquid is let clear for some time, any potassium bi-tartrate, 
which may be formed from small quantities of potassium ad- 
hering to the hydrate, filtered out, and the clear solution 
electrolyzed. There is added to the solution during electrolysis 

* English Patent, Feb. 3, 1887, No. 1751. 
t German Pat. (D. R. P.) 40,626 (1887). 


organic acid — as formic, acetic, citric, oxalic — or, better, ab- 
solute alcohol. 

b. A solution of aluminium nitrate, as far as possible free 
from alkalies and sulphuric acid, is decomposed by a strong 
dynamic current in baths arranged in series, using platinized 
plates as anode and cathode. With a weak current of 0.02 to 
0.05 amperes to a square centimetre, the aluminium separates 
out on the cathode as a deep black deposit, sticking close to 
the copper. The cathode is lifted from the solution, freed from 
small quantities of alumina coating it by gentle rinsing, and 
then the deposit washed off by a strong jet of water. The 
powder obtained is washed further with clear, cold water, par- 
ticularly free from sodium chloride, and dried by gentle heat- 
ing in the air. 

H. C. Bull* proposes to manufacture aluminium alloys by 
using the metal to be alloyed with aluminium as a cathode in a 
bath of aluminium sulphate, the anode being either of alumin- 
ium or of an insoluble substance. When enough aluminium is 
deposited, the cathode is taken out and melted down. 

C. A. Burghardt and W. J. Twining, of Manchester, England, 
have patented the following methods : 

fTo a solution of sodium or potassium aluminate containing 
about 7.2 oz. of aluminium per gallon are added 4 pounds of 
95 per cent, potassium cyanide dissolved in a quart of water, 
and then gradually 2j^ pounds of potassium bi-carbonate. 
The whole is boiled 12 hours and made up to a gallon. The 
bath is used at 175° F. with aluminium or platinum anode and 
carbon or copper cathode. The addition of a little free hydro- 
cyanic acid insures a bright deposit when articles are being 

% Two and one-half kilos of aluminium sulphate in solution is 
precipitated by ammonia, and then re-dissolved by adding \yi 
kilos of caustic soda dissolved in a litre of water; the alumina 

♦English Pat., 10,199 A. (1887). 

t English Pat., July 2, 1887, No. 9389. 

I German Pat. (D. R. P.), 45,020 (1887). 


is thus slightly in excess. The hydrocyanic acid is added until 
a slight precipitate appears. This solution, warmed to 80°, is 
used as a bath from which aluminium is to be deposited. 

*The bath is prepared by dissolving alumina in a solution of 
chloride of copper, and treating further with caustic soda or 
potash for the purpose of causing the aluminium and copper to 
combine together. The precipitate, dissolved in hydrocyanic 
acid and diluted, forms a bath of double cyanide, which when 
electrolyzed deposits an alloy of aluminium and copper. 

Besides the processes so far described, patents have been 
taken out in England by Gerhard and Smith,f Taylor,| and 
Coulson,§ the details of which have not been accessible to the 

Over against these statements of enthusiastic inventors, let 
me place a few extracts from authorities who have given much 
time and attention to the subject of electro-depositing alumin- 
ium from solution without the use of an aluminium anode. 

Sprague|| states his inability to deposit aluminium electric- 
ally from solution. 

Dr. Clemens Winckler IT states that he has spent much time 
in trying all methods so far proposed, and comes to the conclu- 
sion that aluminium cannot be deposited by electricity in the 
wet way. 

Dr. Geo. Gore** although having once proposed a method 
which he said attained this end, yet in his later work on Elec- 
tro-metallurgy does not mention his former proposition, and 
quotes, apparently as coinciding with his own opinion, the 
words of Sprague and Winckler given above. 

* English Pat., Oct. 28, 1887, No. 2602. 

tNo. 16,653 (1884). 

J No. 1991 (1855). 

§ No. 207s (1857). 

II Sprague's Electricity, p. 309. 

t Journal of the Chem. Soc, X. 1 134. 

** Text-book of Electto- metallurgy. 


Dr. S. Mierzinski* states, in 1883, that "the deposition of 
aluminium from an aqueous solution of its salt has not yet been 

Dr. W. Hampef claims to have shown that the electrolysis 
of aqueous aluminous solutions, although frequently patented, 
is not to be expected. From which we would infer that he 
could not testify to it ever having been done. 

Alexander Watt| holds that the electrolytic production of 
aluminium ,from solution is very improbable. He tried acid 
solutions, alkaline solutions, cyanide combinations, etc., under 
most varied conditions, without any result. 

Finally, I will quote from a letter of my good friend Dr. 
Justin D. Lisle, of Springfield, O., who with ample means at 
his disposal, an enthusiasm bred of love tor scientific truth, and 
talent to guide him in his work, has reached the following re- 
sults : " I have tried in almost every conceivable way to de- 
posit it (aluminium) from aqueous solution by electricity, 
using from i pint cells to 60 gallon cells successively ; the cells 
were connected for quantity and for intensity; acid and neutral 
solutions were used; carbon, platinum, and copper electrodes, 
porous cups and diaphragms, were all thoroughly tried, without 
the slightest deposit of metal. In some cases alumina was de- 
posited, which has led me to think that aluminium was pri- 
marily deposited, and owing to the fine state in which it existed 
was promptly oxidized." 

The following details of the operation of the electric plant of 
the Tacony Metal Company, near Philadelphia, under the 
supervision of Mr. J. D. Darling, are all that the writer has 
been able to gather respecting their process. This company 
had the contract to electroplate the exterior ironwork of the 
tower of the Philadelphia Public Buildings, there being alto- 
gether about 50,000 square feet to be plated, on which was to 
be deposited 20,000 pounds of aluminium, or a little over six 

* Die Fabrikation des Aluminiums. 

tChem. Zeit. (Cothen), XI. 935. 

I London Electrical Review, July, 1887. 


ounces to the foot. This would give a smooth coating j\ inch 
thick. The cast-iron work is first given a coating of copper, 
and the aluminium deposited on it, but it is stated that good 
platings have been obtained directly on the cast-iron. 

The largest pieces plated were columns 26 feet long and 3 
feet in diameter. The whole operation of treating such a col- 
umn lasted 1 1 days, while the plant could turn out one every 
4 days. The operations were as follows : 

(i) Boiling 24 hours in a strong solution of caustic soda, 
after which well washed. 

(2) Pickled in dilute sulphuric acid 24 hours, brushed vig- 
orously and washed. 

(3) Copper-plated by immersion 40 hours in a copper 
cyanide(?) solution. The column was then brushed with par- 
affine wax inside, any flaws soldered up, and — 

(4) Copper-plated again in an ordinary copper-plating so- 
lution for 72 hours, until 16 ounces of copper to the square 
foot had been deposited. 

(5 ) Put into the aluminium-plating tank, and 3 ounces of 
aluminium per square foot deposited on it. Time, 72 hours. 
Column then well washed in the sixth tank. 

The dynamo used for operation 3 gives a current of 1000 
amperes at 6 volts tension ; for operation 4, 4000 amperes 
at 2.5 volts tension; for operation 5, 2000 amperes at 8 volts 
tension. In the latter case, the current density at the depos- 
iting surface was 8 amperes per square foot (86 amperes per 
square metre). In this depositing tank, aluminium anodes 4 
feet long, 12 inches wide and }( inch thick were used. Sixty 
of these plates, weighing 35 pounds each, were hung around 
the inside of the tank. 

■ The company will give no information regarding the compo- 
sition of the electrolyte. Concerning the conducting of thS 
operation, Mr. Darling has stated that " Aluminium is more 
difficult to deposit than any of the common metals, because of 
its tendency to re-dissolve after being deposited. I find that 
by using a solution of aluminium salt that has but a slight dis- 


solving effect on aluminium, with a density of current of 8 
amperes per square foot, and with sufficiently high voltage, 
aluminium can be deposited at the rate of i gramme per hour 
per square foot in a reguline state. With much higher cur- 
rents it can be deposited quicker, but will be in a pulverulent 
state, which does not adhere." 


Non-aqueous Electric Processes. 

The above heading was originally written, "The electrolytic 
decomposition of fused aluminium compounds," but I have 
changed it to this form because it is doubtful, in several cases, 
whether electrolytic decomposition really plays much of a role 
in the isolation of the aluminium. In some processes a pow- 
erful current is used and largely converted into heat, producing 
temperatures never before used in industrial operations. In 
such cases, chemical reactions, rendered possible by the very 
high temperature, come into play and may even be the prin- 
cipal means of reducing the aluminium compound. It has been 
suggested that such processes should be classed apart as 
"electro-thermal methods;" but such a division does not ap- 
pear satisfactory to me, because in most cases we cannot say 
positively that electrolysis does not play some part, and in 
others it is simply impossible to decide which of the two agents 
of reduction, electrolysis or chemical affinity, is the most active. 
It has been suggested also that those processes in which high 
voltage is used cannot be electrolytic, but must be electro-ther- 
mal ; but such a division is not logical. If a bath is taking a 
high voltage to work it, it may be solely from poor conductivity 
in the electrolyte, and not because an arc is formed in the bath, 
and true electrolysis may be going on to the full extent of the 
power of the current. If we knew definitely, for each process, 
which action played the principal part in the reduction, we 
might classify the processes, but such a classification would not 
be of much use. If it was absolutely necessary to make some 


such division, I should consider as electro-thermal those pro- 
cesses in which the parts of the apparatus used are not ar- 
ranged as they should be to secure the maximum electrolytic 
decomposition. We could thus, in such cases, say at once, 
from a consideration of the apparatus and its manner of work- 
ing, that electrolysis was evidently not contemplated in its con- 
struction and could only play a minor part in its operation. 

In conclusion, I regard it impossible to make an exact clas- 
sification of these processes such as would be of any value to 
the metallurgist or aid the general reader to a better under- 
standing of them, and I therefore take them up in chronolog- 
ical order. 

Davy's Experiment (1810). 

Sir Humphry Davy, in his Brompton Lecture before the 
Royal Philosophical Society,* described the following attempt 
to decompose alumina and obtain the metal of this earth. He 
connected an iron wire with the negative pole of a battery con- 
sisting of 1000 double plates. The wire was heated to white- 
ness and then fused in contact with some moistened alumina, 
the operation being performed in an atmosphere of hydrogen. 
The iron became brittle, whiter, and on being dissolved in acid 
gave a solution from which was precipitated alumina, identical 
with that used. 

It is evident that the alumina had been reduced by the 
chemical action of the hydrogen gas, a reaction which we now 
know takes place below the fusing point of alumina, 2200°, and 
probably begins as low as 1700°. 

Duvivier's Experiment (1854). 

M. Duvivier f states that by passing an electric current from 
eighty Bunsen cells through a small piece of laminated disthene 
between two carbon points, the disthene melted entirely in two 
or three minutes ; the elements which composed it were partly 

* Philosophical Transactions, 1810. 
fThe Chemist, Aug., 1854. 


disunited by the power of the electric current, and some alu- 
minium freed from its oxygen. Several globules of the metal 
separated, one of which was as white and as hard as silver. 

Since disthene is aluminium silicate, formula Al^Os.SiOj, it 
is most likely that the metal obtained was highly siliceous 
aluminium, since complete reduction of this compound would 
give 54 parts of aluminium to 28 parts of silicon. The highly- 
heated carbon particles of the electric arc were most probably 
the reducing agent. 

Bunsen's and Deville's Methods (1854). 

A method of decomposing aluminium-sodium chloride by the 
battery was discovered simultaneously by Deville in France and 
Bunsen in Germany, in 1854, and is nothing else but an appli- 
cation of the process already announced by Bunsen of decom- 
posing magnesium chloride by the battery. Deville gives the 
more minute account, and we therefore quote his description of 
the process. 

f " It appears to me impossible to obtain aluminium by the 
battery in aqueous solutions. I should believe this to be an 
absolute impossibility if the brilliant experiments of M. Bunsen 
in the preparation of barium, chromium and manganese did not 
shake my convictions. Still I must say that all the processes 
of this description which have recently been published for the 
preparation of aluminium have failed to give me any results. 
Every one knows the elegant process by means of which M. 
Bunsen has lately produced magnesium, decomposing fused mag- 
nesium chloride by an electric current. The illustrious profes- 
sor at Heidelberg has opened up a method which may lead to 
very interesting results. However, the battery cannot be used 
for decomposing aluminium chloride directly, which does not 
melt, but volatilizes at a low temperature ; it is, therefore, neces- 
sary to use some other material which is fusible and in which 

*Tbe Chemist, Aug. 1854. 

t Ann. de Chem. et de Phys. [3], 46, 452; Deville's " de rAluminium." 


aluminium alone will be displaced by the current. I have found 
this salt in the double chloride of aluminium and sodium, which 
melts towards 185° C, is fixed at a somewhat high temperature, 
although volatile below the fusing point of aluminium, and thus 
unites all the desirable conditions. 

" I put some of this double chloride into a porcelain crucible 
separated imperfectly into two compartments by a thin leaf of 
porcelain, and decomposed it by means of a battery of five ele- 
ments and carbon electrodes. The crucible was heated more 
and more as the operation progressed, for the contents became 
less and less fusible, but the heat was not carried past the melt- 
ing point of aluminium. Arrived at this point, after having 
lifted out the diaphragm and electrodes, I heated the crucible 
to bright redness, and found at the bottom a button of alumin- 
ium, which was flattened out and shown to the Academy in the 
S6ance of March 20, 1854. The button was accompanied by a 
considerable quantity of carbon, which prevented the union of 
a considerable mass of shot-metal. This carbon came from the 
disintegration of the very dense gas-retort carbon electrodes; in 
fact, the positive electrode was entirely eaten away in spite of 
its considerable thickness. It was evident, then, that this appa- 
ratus, although similar to that adopted by Bunsen for manufac- 
turing magnesium, would not suit here, and the following is the 
process which after many experiments I hold as best. 

" To prepare the bath for decomposition, I heat a mixture of 
2 parts aluminium chloride and i part sodium chloride, dry and 
pulverized, to about 200° in a porcelain capsule. They com- 
bine with disengagement of heat, and the resulting bath is very 
fluid. The apparatus which I use for the decomposition com- 
prises a glazed porcelain crucible, which as a precaution is 
placed inside a larger one of clay. The whole is covered by a 
porcelain cover pierced by a slit to give passage to a large thick 
leaf of platinum, which serves as the negative electrode ; the lid 
has also a hole through which is introduced, fitting closely, a 
well-dried porous cylinder, the bottom of which is kept at some 
distance from the inside of the porcelain crucible. This porous 



vessel encloses a pencil of retort carbon, which serves as the 
positive electrode. Melted double chloride is poured into the 
porous jar and into the crucible so as to stand at the same 
height in both vessels ; the whole is heated just enough to keep 
the bath in fusion, and there is passed through it the current 
from several Bunsen cells, two cells being strictly sufficient. 
The annexed diagram shows the crucibles in section. 

Fig. 28. 

"The aluminium deposits with some sodium chloride on the 
platinum leaf; the chlorine, with a little aluminium chloride, is 
disengaged in the porous jar and forms white fumes, which are 
prevented from rising by throwing into the jar from time to 
time some dry, pulverized sodium chloride. To collect the 
aluminium, the platinum leaf is removed when sufficiently 
charged with the saline and metallic deposit; after letting it 
cool, the deposit is rubbed off and the leaf placed in its former 
position. The material thus detached, melted in a porcelain 
crucible, and after cooling washed with water, yields a gray, 
metallic powder, which by melting several times under a layer 
of the double chloride is reunited into a button." 

Bunsen* adopted a similar arrangement. The porcelain cru- 
* Pogg. Annalen, 97, 648. 


cible containing the bath of aluminium-sodium chloride kept in 
fusion was divided into two compartments in its upper part by a 
partition, in order to separate the chlorine liberated from the alu- 
minium reduced. He made the two electrodes of retort carbon. 
To reunite the pulverulent aluminium, Bunsen melted it in a bath 
of the double chloride, continually throwing in enough sodium 
chloride to keep the temperature of the bath about the fusing 
point of silver. 

As we have seen, Deville, without being acquainted with Bun- 
sen's investigations, employed the same arrangement, but he 
abandoned it because the retort carbon slowly disintegrated in 
the bath, and a considerable quantity of double chloride was lost 
by the higher heat necessary to reunite the globules of alumin- 
ium after the electrolysis. Deville also observed that by working 
at a higher temperature, as Bunsen had done, he obtained purer 
metal, but in less quantity. The effect of the high heat is that 
silicon chloride is formed and volatilizes, and the iron which 
would have been reduced with the aluminium is transformed 
into ferrous chloride by the aluminium chloride, and thus the 
aluminium is purified of silicon and iron. 

Plating aluminium on copper. — The same bath of double chlor- 
ide of aluminium and sodium may be used for plating alumin- 
ium in particular on copper, on which Capt. Caron experimented 
with Deville. Deville says : " To succeed well, it is necessary to 
use a bath of double chloride which has been entirely purified 
from foreign metallic matter by the action of the battery itself. 
When aluminium is being deposited at the negative pole, the first 
portions of metal obtained are always brittle, the impurities in 
the bath being removed in the first metal thrown down; so, 
when the metal deposited appears pure, the piece of copper to 
be plated is attached to this pole and a bar of pure aluminium 
to the positive pole. However, a compact mixture of carbon 
and alumina can be used instead of the aluminium anode, which 
acts similarly to it and keeps the composition of the bath con- 
stant. The temperature ought to be kept a little lower than the 
fusing point of aluminium. The deposit takes place readily 


and is very adherent, but it is difficult to prevent it being im- 
pregnated with double chloride, which attacks it the moment 
the piece is washed. The washing ought to be done in a large 
quantity of water Cryolite might equally as well be used for 
this operation, but its fiisibility should be increased by mixing 
with it a little double chloride of aluminium and sodium and 
some potassium chloride." 

During a recent law-suit for infringement of the Hall process, 
it was held by the defendants that Deville electrolyzed a fluoride 
bath, (cryolite) in which alumina was present and therefore 
would be dissolved. It appears quite evident, however, that 
Deville did not use cryolite, because it would have destroyed 
the porcelain crucibles and necessitated the use of special ap- 
paratus which Deville does not even hint at. It is, moreover, 
true that when alumina is mixed with carbon the latter prevents 
the cryoHte from wetting or coming into actual contact with it. 
In an experiment by the writer, a large cylinder of carbon into 
which had been incorporated over 50 per cent, of alumina 
was kept in molten cryolite for several hours, and on being 
taken out showed not the slightest sign of corrosion. Judges 
Taft and Ricks of Ohio very properly ruled that Deville's ex- 
periments did not in any way disclose or operate on the princi- 
ple utilized by Mr. Hall. 

Le Chatellier's Method (1861). 

The subject of this patent* was the decomposition of the 
fused double chloride of aluminium and sodium, with the par- 
ticular object of coating or plating other metals, the articles 
being attached to the negative pole. About the only novelty 
claimed in this patent was the use of a mixture of alumina and 
carbon for the anode, but we see from the previous paragraph 
that this was suggested by Deville several years before ; the 
only real improvement was the placing of this anode inside a 
porous cup, in order to prevent the disintegrated carbon from 

* English Patent, 1861, No. 1214. 


falling into the bath. This was really only the Deville process, 
patented in England. 

Monckton's Patent (1862). 

Monckton* proposes to pass an electric current through a 
reduction chamber, and in this way to raise the temperature 
to such a point that alumina will be reduced by the carbon 
present. We clearly see in this the germ of several more- 
recently patented processes. 

Gaudin's Process (1869). 

Gaudin f reduces aluminium by a process to which he ap- 
plies the somewhat doubtful title of economic. He melts to- 
gether equal parts of cryolite and sodium chloride, and traverses 
the fused mass by a galvanic current. Fluorine is evolved at 
the positive pole, while aluminium accumulates at the negative. 

The great difficulties in such a process are the obtaining of 
suitable vessels to withstand the corrosion of the bath, the great 
inconvenience of the escaping fluoride gas, and the increase of 
resistance due to the change of composition of the bath. It 
would need several additional patents to render this process 
operative, without mentioning economy. 

Kagensbusch' s Process (1872). 

Kagensbusch,! of Leeds, proposes to melt clay with fluxes, 
then adding zinc or a like metal to pass an electric current 
through the fused mass, isolating an alloy of aluminium and 
the metal, from which the foreign metal may be removed by 
distillation, sublimation, or cupellation. 

If the patentee had used pure alumina instead of clay, he 
might have produced some aluminium by his process, but the 
silicon in the clay would get into the zinc in larger proportion 

* English Patent, 1862, No. 264. 
t Moniteur Scientifique, xi., 62. 
X English Patent, 1872, No. 481 1. 


than the aluminium, rendering the succeeding operations for 
obtaining pure aluminium impossible. 

Berthaut's Proposition (1879). 

Up to this time all the proposed electric processes were 
confined to the use of a galvanic current, the cost of obtaining 
which was a summary bar to all ideas of economical produc- 
tion. About this period dynamo-electric machines were being 
introduced into metallurgical practice, and Berthaut is the first 
we can find who proposes their use in producing aluminium. 
The process which he patented * is otherwise almost identical 
with Le Chatellier's, 

Grdtzel's Process (1883). 

This process f has little claim to originality, except in the 
details of the apparatus. A dynamo-electric current is used, 
the electrolyte is fused cryolite or double chloride of aluminium 
and sodium, and the anodes are of pressed carbon and alumina 
— none of which points are new. However, the use of melting 
pots of porcelain, alumina, or aluminium, and making them the 
negative electrode, are points in which innovations are made. 

In a furnace are put two to five pots, according to the power 
of the dynamo used, each pot having a separate grate. The 
pots are preferably of metal, cast-steel is used, and form the 
negative electrodes. The positive electrode, K (Fig. 29), can 
be made of a mixture of anhydrous alumina and carbon pressed 
into shape and ignited. A mixture of alumina and gas-tar 
answers very well ; or it can even be made of gas-tar and gas- 
retort carbon. During the operation little pieces of carbon fall 
from it and would contaminate the bath, but are kept from 
doing so by the mantle G. This isolated vessel, G, is perforated 
around the lower part at g, so that the molten electrolyte may 
circulate through. The tube O' conducts reducing gas into the 

* English Patent, 1879, No. 4087. 

t German Patent (D. R. P.), No. 26962 (1883). 



crucible, which leaves by the tube 0\ This reducing atmos- 
phere is important, in order to protect from burning any metal 
rising to the surface of the bath. The chlorine set free at the 
electrode, K, partly combines with the alumina in it, regenera- 
ting the bath, but some escapes, and, collecting in the upper 
part of the surrounding mantle, G, is led away by a tube con- 

FlG. 29. 

necting with it. Instead of making the electrode K of carbon 
and alumina, it may simply be of carbon, and then plates of 
pressed alumina and carbon are placed in the bath close to the 
electrode, K, but not connected with it. Also, in place of 
making the crucible of metal and connecting it with the nega- 
tive pole, it may be made of a non-conducting material, clay or 
the like, and a metallic electrode — as, for instance, of alu- 
minium — plunged into the bath. 

In a later patent,* Gratzel states that the bath is decomposed 

* English Patent; 14325, Nov. 23, 1885. U. S. Patent, 362441, May 3, 1887. 


by a current of comparatively low tension if magnesium chloride 
be present ; the chlorides of barium, strontium or calcium act 

Prof. F. Fischer* maintains as impracticable the use of plates 
of pressed alumina and carbon, which can, further, only be 
operative when they are made the positive electrode, and then 
their electric resistance is too great. The incorporation into 
them of copper filings, saturation with mercury, etc., give no 
more practical results. There are also volatilized at the anodes 
considerable quantities of aluminium chloride, varying in 
amount with the strength of the current. 

Large works were erected near Bremen by the Aluminium 
und Magnesiumfabrik Pt. Gratzel, zu Hemelingen, in which 
this process was installed. License was also granted to the 
large chemical works of Schering, at Berlin, to operate it. R. 
Biedermann, in commenting on the process in i886,f stated 
that the results obtained so far were not fully satisfactory, but 
the difficulties which had been met were of a kind which would 
certainly be overcome. They were principally in the polariza- 
tion of the cathode, by which a large part of the current was 
neutralized. By using proper depolarizing substances this dif- 
ficulty would be removed. The utilization of the chlorine 
evolved would also very much decrease the expenses. A more 
suitable slag, which collected the aluminium together better, 
was also desirable. Finally, the metal produced was somewhat 
impure, taking up iron from the iron pots and silicon from the 
clay ones, to obviate which Biedermann recommended the use 
of lime or magnesia vessels. 

Prof. Fischer, as we have seen, maintained the uselessness of 
Gratzel's patent claims, and his later expression of this opinion 
in 1887 drew a reply from A. Saarburger, Jdirectorof the works 
at Hemehngen, to the effect that since October, 1887, they had 

* Wagner's Jahresbericht, 1884, p. 1319; 1887, p. 376. 

fKerl und Stohman, 4th ed., p. 725. 

{ Verein der Deutsche Ingenieure, Jan. 26, 1889. 


abandoned the Gratzel process, and were making aluminium at 
present by methods devised by Herr Saarburger; in conse- 
quence of which fact the directors of the company decided in 
January, 1888, to drop the addition Pt. Gratzel from the firm 
name. The methods then in use at Hemelingen were kept secret, 
but the author was informed by a friend in Hamburg that they 
were using a modified Deville sodium process. Herr Saarbur- 
ger informed me in October, 1888, that they were producing 
pure aluminiuin at the rate of 12 tons a year, besides a large 
quantity sold in alloys. An attractive pamphlet issued by this 
firm set forth precautions to be used in making aluminium 
alloys, together with a digest of their most important properties, 
which we shall have occasion to quote from later in considering 
those alloys. 

The sharp decline in the price of aluminium in 1890 forced 
this company out of the aluminium business. 

Cowles Bros! Process. 

Messrs. E. H. and A. H. Cowles patented in the United 
States and Europe * an electric furnace and its application for 
producing aluminium. Their patent claims " reducing an alu- 
minium compound in company with a metal in presence of car- 
bon in a furnace heated by electricity ; the alloy of aluminium 
and the metal formed being further treated to separate out the 
aluminium." The history of the development of this process 
having already been sketched, we will proceed to describe the 
details of its operation. The first public description was given 
in two papers, one read before the American Association for the 
Advancement of Science f by Prof. Chas. F. Mabery of the Case 
School of Applied Science, Cleveland, the other before the 
American Institute of Mining Engineers % by Dr. T. Sterry 
Hunt, of Montreal. 

* U. S. Patents 324658, 324659, Aug. 18, 1885; English Patent 9781, same date; 
German Patent 33672. 

t Ann Arbor Meeting, Aug. 28, 1885. 
X Halifax Meeting, Sept. 16, 1885. 


Prof. Mabery said in his paper : " Some time since, the Messrs. 
Cowles conceived the idea of obtaining a continuous high tem- 
perature on an extended scale by introducing into the path of 
an electric current some material that would afford the requi- 
site resistance, thereby producing a corresponding increase in 
the temperature. After numerous experiments, coarsely pul- 
verized carbon was selected as the best means for maintaining 
an invariable resistance, and at the same time as the most 
available substance for the reduction of oxides. When this 
material, mixed with the oxide to be reduced, was made a part 
of the electric circuit, inclosed in a fire-clay retort, and sub- 
jected to the action of a current from a powerful dynamo, not 
only was the oxide reduced, but the temperature increased to 
such an extent that the whole interior of the retort fused com- 
pletely. In other experiments lumps of lime, sand, and corun- 
dum were fused, with a reduction of the corresponding metal ; 
on cooling, the lime formed large, well-defined crystals, the 
corundum beautiful red-green and blue octahedral crystals. 
Following up these results, it was soon found that the intense 
heat thus produced could be utilized for the reduction of oxides 
in large quantities, and experiments were next tried on a large 
scale with the current from a fifty horse-power dyamo. For 
the protection of the walls of the furnace, which were of fire- 
brick, a mixture of ore and coarsely pulverized gas-carbon was 
made a central core, and was surrounded on the side and bot- 
tom by fine charcoal, the current following the lesser resistance 
of the core from carbon electrodes inserted in the ends of the 
furnace in contact with the core. The furnace was charged by 
first filling it with charcoal, making a trough in the centre, and 
filling this with the ore mixture, the whole being covered with 
a layer of coarse charcoal. The furnace was closed on top 
with fire-brick slabs containing tivo or three holes for the 
escape of the gaseous products of the reduction, and the whole 
furnace was made air-tight by luting with fire-clay. Within a 
few minutes after starting the dynamo, a stream of carbonic 
oxide issued through the openings, burning usually with a flame 


eighteen inches high. The time required for complete reduc- 
tion was ordinarily about an hour. Experience has already 
shown that aluminium, silicon, boron, manganese, sodium and 
potassium can be reduced from their oxides with ease. In fact, 
there is no oxide that can withstand the temperature attainable 
in this furnace. Charcoal is changed to graphite; does this 
indicate fusion? As to what can be accomplished by convert- 
ing enormous electrical energy into heat within narrow limits, 
it can only be said that it opens the way into an extensive field 
of pure and applied chemistry. It is not difficult to conceive 
of temperature limited only by the power of carbon to resist 

" Since the motive power is the chief expense in accomplish- 
ing reductions by this method, its commercial success is 
closely connected with obtaining power cheaply. Realizing 
the importance of this point, Messrs. Cowles have purchased at 
Lockport, N. Y., a water-power where they can utilize I200 
horse-power. An important feature in the use of these furnaces 
from a commercial standpoint is the slight technical skill re- 
quired in their manipulation. The four furnaces operated in 
the experimental laboratory at Cleveland are in charge of two 
young men, who six months ago knew absolutely nothing 
of electricity. The products at present manufactured are 
the various grades of aluminium bronze, made from a rich 
furnace product obtained by adding copper to the charge of 
ore. Aluminium silver is also made ; and a boron bronze may 
be prepared by the reduction of boracic acid in contact with 
copper, while silicon bronze is made by reducing silica in con- 
tact with copper. As commercial results may be mentioned 
the production in the experimental laboratory, which averages 
50 lbs. of 10 per cent, aluminium bronze daily, which can be 
supplied to the trade in large quantities on the basis of $5 per 
lb. for the aluminium contained, the lowest market quotation of 
aluminium being now $15 per lb." 

Dr. Hunt stated further that if the mixture consisted of alu- 
mina and carbon only, the reduced metal volatilized, part es- 


caping into the air and burning to alumina, part condensing in 
the upper layer of charcoal, affording thus crystalline masses of 
nearly pure aluminium and yellow crystals supposed to be a 
compound of aluminium with carbon. Great loss was met in 
collecting this divided metal into an ingot, so that only small 
quantities were really obtained. To gather all the aluminium 
together, a metal such as copper was added, thus producing an 
alloy with 15 to 20 per cent, of aluminium ; on substituting 
this alloy for pure copper in another operation, an alloy with 
over 30 per cent, of aluminium was obtained. 

Dr. Hunt, in a later paper,* stated that pure aluminium has 
been obtained in this process by first producing in the furnace 
an alloy of aluminium and tin, then melting this with lead, 
when the latter takes up the tin and sinks with it beneath 
the aluminium. He also stated that in the early experiments a 
dynamo driven by a 30 horse-power engine yielded a daily out- 
put of 50 lbs. of 10 per cent, aluminium bronze, but with a 
larger machine the output was proportionally much greater. 
In the latest practice, one-half cent per horse-power per hour 
is said to cover the expense of working, making the 10 per 
cent, bronze cost about 5 cents per lb. over the copper used. 

Various shapes of furnaces have been used by the Cowles 
Bros., the first described being a rectangular box, lined with 
carbon, with the electrodes passing through the ends. Al- 
though two other forms have been patented, we understand 
that the kind now used, aad which is described at length in Mr. 
Thompson's paper, is also of the oblong, horizontal style. 
Chas. S. Bradley and Francis B. Crocker, of New York, patented 
and assigned to the Cowles Electric Smelting Company,f the 
use of a retort, composed of conducting material, surrounded 
by a substance which is a poor conductor of heat, and having 
inside a mixture of charcoal and the ore to be heated. Electric 
connection being made with the ends of the retort, the walls of 

♦National Academy of Science, Washington Meeting, April 30, 1886. 
tU. S. Patent 335499, Feb. j., 1886. 


the retort and the material in it are included in the circuit and 
constitute the greater part of the resistance. The retort may 
be stoppered at each end during the operation, and the heating 
thus performed in a reducing atmosphere. Mr. A. H. Cowles 
devised a style of furnace adapted for continuous working and 
utilizing the full current of a dynamo of the largest size.f The 
electrodes are tube-shaped and placed vertically. The positive 
pole is above, and is surmounted by a funnel in which the 
mixture for reduction is placed. The regular delivery of the 
mixture is facilitated by a carbon rod, passing through the 
cover of the funnel, which is serrated on the end and can be 
worked up and down. The melted alloy produced, with any 
slag, passes down through the negative electrode. The dis- 
tance between the poles can be regulated by moving the upper 
one, and the whole is inclosed in a fire-brick chamber. The 
space between the electrodes and the walls is filled with an iso- 
lating material, which is compact around the lower electrode 
but coarse grained around the upper to facilitate the escape of 
the gases produced. The chamber is tightly closed excepting 
a small tube for the escape of gas. 

A very complete description of the Cowles process was given 
by Mr, W. P. Thompson (agent for the Cowles Co. in England) 
in a paper read before the Liverpool Section of the Society of 
Chemical Industry. f He describes the process as then carried 
on in Lockport, the dynamo used being a large Brush machine 
weighing 2J^ tons and consuming about lOO horse-power in 
being driven at 900 revolutions per minute. 

" Conduction of the current of the large dynamo to the furnace 
and back is accompHshed by a complete metallic circuit, except 
where it is broken by the interposition of the carbon electrodes 
and the mass of pulverized carbon in which the reduction takes 
place. The circuit is of 13 copper wires, each 0.3 inch in dia- 
meter. There is likewise in the circuit an ampere meter, or am- 

* English Patent 4664 (1887). 
.f Journal of the Society of Chemical Industry, April 29, 1886. 



meter, through whose helix the whole current flows, indicating 
the total strength of the current being used. This is an import- 
ant element in the management of the furnace, for, by the posi- 
tion of the finger on the dial, the furnace attendant can tell to a 
nicety what is being done by the current in the furnace. Be- 
tween the ammeter and the furnace is a resistance coil of Ger- 
man silver kept in water, throwing more or less resistance into 
the circuit as desired. This is a safety appliance used in 
changing the current from one furnace to another, or to choke 
off the current before breaking it by a switch. 

"The furnace (see Figs. 30, 31, 32) is simply a rectangular 

Fig. 30. 

box. A, one foot wide, five feet long inside, and fifteen inches 
deep, made of fire-brick. From the opposite ends through the 
pipes BB the two electrodes CC pass. The electrodes are im- 
mense electric-light carbons three inches in diameter and thirty 

Fig. 31. 


inches long. If larger electrodes are required, a series this size 
must be used instead, as so far all attempts to make larger car- 
bons that will not disintegrate on becoming incandescent have 
failed. The ends of the carbons are placed within a few inches 



of each other in the middle of the furnace, and the resistance 
coil and ammeter are placed in the circuit. The ammeter regis- 
ters 50 to 2000 amperes. These connections made, the furnace 
is ready for charging. 

" The walls of the furnace must first be protected, or the in- 
tense heat would melt the fire-brick. The question arose, what 
would be the best substance to line the walls. Finely powdered 
charcoal is a poor conductor of electricity, is considered infusi- 
ble and the best non-conductor of heat of all solids. From 
these properties it would seem the best material. As long as 

Fig. 32. 


air is excluded it will not burn. But it is found that after using 
pure charcoal a few times it becomes valueless ; it retains its 
woody structure, as is shown in larger pieces, but is changed 
to graphite, a good conductor of electricity, and thereby tends 
to diffuse the current through the lining, heating it and the walls. 
The fine charcoal is therefore washed in a solution of Hme- 
water, and after drying each particle is insulated by a fine 
coating of lime. The bottom of the furnace is now filled with 
this lining about two or three inches deep. A sheet-iron gauge 
is then placed along the sides of the electrodes, leaving about 
two inches between them and the side walls, in which space 
more of the charcoal is placed. The charge E, consisting of 
about 25 pounds of alumina, iii its native form as corundum, 12 


pounds of charcoal and carbon, and 50 pounds of granulated 
copper, is now placed within the gauge and spread around the 
electrodes to within a foot of each end of the furnace. For 
making iron alloy where silicon also is not harmful, bauxite or 
various clays containing iron and silica may be used instead of 
the pure alumina or corundum. In place of granulated cop- 
per, a series of short copper wires or bars can be placed parallel 
to each other and transverse to the furnace, among the alumina 
and carbon, it being found that where grains are used they 
sometimes fuse together in such a way as to short-circuit the 
current. After this, a bed of charcoal, F, the granules of which 
vary in size from a chestnut to a hickory, is spread over all, 
and the gauge drawn out. This coarse bed of charcoal above 
the charge allows free escape of the carbonic oxide generated in 
the reduction. The charge being in place, an iron top, G, 
lined with fire-brick, is placed over the whole furnace and the 
crevices luted to prevent access of air. The brick of the walls 
insulate the cover from the current. 

" Now that the furnace is charged, and the cover luted down, 
it is started. The ends of the electrodes were in the beginning 
placed close together, as shown in the longitudinal section, and 
for this cause the internal resistance of the furnace may be too 
low for the dynamo, and cause a short circuit. The operator, 
therefore, puts sufficient resistance into the circuit, and by 
watching the ammeter, and now and then moving one of the 
electrodes out a trifle, he can prevent undue short-circuiting in 
the beginning of the operation. In about ten minutes, the cop- 
per between the electrodes has been melted and the latter are 
moved far enough apart so that the current becomes steady. 
The current is now increased till 1300 amperes are going 
through, driven by 50 volts. Carbonic oxide has already com- 
menced to escape through the two orifices in the top, where it 
burns with a white flame. By slight movements outward of the 
electrodes during the coming five hours, the internal resistance 
in the furnace is kept constant, and at the same time all the 
different parts of the charge are brought in turn into the zone 


of reduction. At the close of the run the electrodes are in the 
position shown in the plan, the furnace is shut down by placing 
a resistance in the circuit, and then the current is switched into 
another furnace charged in a similar manner. It is found that 
the product is larger if the carbons are inclined at angles of 30° 
to the horizontal plane, as in Fig. 33. 

Fig. 33- 


" This regulating of the furnace by hand is rather costly and 
unsatisfactory. Several experiments have therefore been tried 
to make it self-regulating, and on January 26, 1886, a British 
patent was applied for by Cowles Bros, covering an arrange- 
ment for operating the electrodes by means of a shunt circuit, 
electro-magnet, and vibrating armature. Moreover, if the elec- 
trodes were drawn back and exposed to the air in their highly 
heated state, they would be rapidly wasted away. To obviate 
this, Messrs. Cowles placed what may be called a stufifiing-box 
around them, consisting of a copper box filled with copper shot. 
The wires are attached to the boxes instead of the electrodes. 
The hot electrodes as they emerge from the furnace first en- 
counter the shot, which rapidly carry off the heat, and by the 
time they emerge from the box they are too cool to be oxidized 
by contact with the air. 

" Ninety horse-power have been pumped into the furnace for 
five hours. At the beginning of the operation the copper first 
melted in the centre of the furnace. There was no escape for 
the heat continually generated, and the temperature increased 
until the refractory corundum melted, and being surrounded on 
all sides by carbon, gave up its oxygen. This oxygen, uniting 
with the carbon to form carbonic oxide, has generated heat 


which certainly aids in the process. The copper has had noth- 
ing to do with the reaction, as it will take place in its absence. 
Whether the reaction is due to the intense heat or to electric 
action it is difficult to say. If it be electric, it is Messrs. Cowles' 
impression that we have here a case where electrolysis can be 
accomplished by an alternating current, although it has not 
been tried as yet. Were the copper absent, the aluminium set 
free would now absorb carbon, and become a yellow crystal- 
line carbide of aluminium ; but, instead of that, the copper has 
become a boiling, seething mass, and the bubblings of its vapors 
may distinctly be heard. The vapors probably rise an inch or 
two, condense, and fall back, carrying with them the freed alu- 
minium. This continues until the current is taken ofif the fur- 
nace, when we have the copper charged with 1 5 to 30 per cent., 
and in some cases as high as 40 per cent, of its weight of alu- 
minium, and a little silicon. After cooling the furnace this 
rich alloy is removed. A valuable property of the fine charcoal 
is that the metal does not spread and run through the inter- 
stices, but remains as a liquid mass, surrounded below and on 
the sides by fine charcoal, which sustains it just as flour or 
other fine dust will sustain drops of water for considerable 
periods, without allowing them to sink in. The alloy is white 
and brittle. This metal is then melted in an ordinary crucible 
furnace, poured into large ingots, the amount of aluminium in 
it determined by analysis, again melted, and the requisite 
amount of copper added to make the bronze desired. 

"Two runs produce in ten hours' average work 100 pounds 
of white metal, from which it is estimated that Cowles Bros., at 
Lockport are producing aluminium in its alloys at a cost of 
about 40 cents per lb. The Cowles Company will shortly have 
1200 horse-power furnaces. With a larger furnace there is no 
reason why it should not be made to run continuously, like the 
ordinary blast furnace. 

" In place of the copper any non-volatile metal may be used 
as a condenser to unite with any metal it may be desired to 
reduce, provided, of course, that the two metals are of such a 


nature that they will unite at this high temperature. In this 
way aluminium may be alloyed with iron, nickel, silver, tin or 
cobalt. Messrs. Cowles have made alloys containing 50 alu- 
minium to 50 of iron, 30 aluminium to 70 of copper, and 25 
aluminium to 75 of nickel. Silicon or boron or other rare 
metals may be combined in the same way, or tertiary alloys 
may be produced ; as, for instance, where fire- clay is reduced 
in presence of copper we obtain an alloy of aluminium, silicon 
and copper." 

Soon after Mr. Thompson's description, the plant at Lock- 
port was increased by the addition of the largest dynamo up to 
that time constructed, built by the Brush Electric Company, 
and dubbed the " Colossus." This machine weighs almost ten 
tons, and when driven at 423 revolutions per minute it pro- 
duced a useful current of 3400 amperes at a tension of 68 volts, 
or at 405 revolutions produced a current of 3200 amperes with 
83 volts electro-motive force, indicating 249,000 Watts or 334 
electric horse-power. The steam engine was, in the latter case, 
developing nearly 400 horse-power, and could not supply 
more ; it was judged that the dynamo could have been driven 
to 300,000 Watts with safety. The first run with this machine 
was made in September, 1886. The furnaces used for this cur- 
rent are of the same style as that described by Mr. Thompson, 
but are larger, the charge being 60 lbs. of corundum, 60 lbs. of 
granulated copper, 30 lbs. of coarse charcoal besides the pulver- 
ized lime-coated charcoal used in packing. The operation of 
reducing this charge takes about two hours. As soon as the 
operation is finished the current is switched off into another 
furnace prepared and charged, so that the dynamo is kept work- 
ing continuously. In 1888, the Cowles Company had two of 
these large dynamos in operation, and eight furnaces in use. 
With two-hour runs a furnace is tapped every hour, producing 
about 80 lbs. of bronze averaging 18 per cent, of aluminium. 
The capacity of the plant is, therefore, about i}i tons of 10 
per cent, bronze per day. Their alloys were sold in 1890 on 
the basis of $2.50 per lb. for the contained aluminium. 



The Cowles Syndicate Company, of England, located at 
Stoke-on-Trent, have set up a large plant at Milton, where, pro- 
fiting by the experience of the parent concern in America, still 
larger electric currents are used, these being found more eco- 
nomical. The dynamo in use at this works was built by Cromp- 
ton, and supplies a current of 5000-6000 amperes at 50 to 60 
volts. There are two furnace-rooms, each containing six fur- 
naces, aluminium and silicon bronze being produced in one 
room, and ferro-aluminium in the other. The furnaces used 
measure 60 by 20 by 36 inches, inside dimensions, and differ 
from those previously shown by having the electrodes inclined 
at an angle of about 30°. (See Figs. 33, 34). The electrodes 


used are formed by bundling together 9 carbon rods, each 2^^ 
inches in diameter, each electrode weighing 20 lbs. More re- 
cently larger carbons have been obtained, 3 inches in diameter, 
and an electrode formed of five of these weighs 36 lbs. The 
furnace is charge^ as previously described. 

The current is started at 3000 amperes, gradually increasing 
to 5000 during the first half hour, and then keeping steady 
until the run is ended, which is about one and a half hours 
from starting. The product of each run is about 100 lbs. of 
raw bronze containing 15 to 20 per cent, of aluminium. The 
return is said to average i lb. of contained aluminium per 18 


electric horse-power hours, or ij^ lbs. per electric horse- 
power per day. The works produce about 200 lbs. of alumin- 
ium contained in alloys per day. The raw bronze is stacked 
until several runs have accumulated, then a large batch is melted 
at once in a reverberatory furnace, refined and diluted to the 
proportion of aluminium required by adding pure copper. 

The Cowles Company, both in England and America, pro- 
duce six standard grades of bronze as follows : 

'Special" A. 








cent, of aluminium 


( it 


it It 


t ti 

t it 
i ti 

Their ferro-aluminium is sold with usually 5 to 7 per cent. 
of aluminium, but 10, 13, and 15 per cent, is furnished if 
asked for. 

Products of the Cowles furnace. — Dr. W.. Hampe obtained the 
following results on analyzing a sample of Cowles Bros.' 10 per 
cent, bronze : 

Copper 90.058 

Aluminium 8.236 

Silicon 1.596 

Carbon. 0.104 

Magnesium 0.019 

Iron trace. 


A sample of 10 per cent, bronze, made in the early part of 
1886, and analyzed in the laboratory of the Stevens Institute, 
showed — 

Copper 88.0 

Aluminium 6.3 

Silicon 6.5 

but it is evident that the percentage of silicon has since then 
been lowered. 


The ferro-aluminium used by Mr. Keep in his tests on cast- 
iron was furnished by the Cowles Company, and analyzed — 

Aluminium I ^4^ 

Silicon , 3'86 

A sample shipped to England in December, 1886, contained — 

Iron 86.69 

Combined carbon I.oi 

Graphitic carbon 1.9' 

Total 2.92 

Silicon 2.40 

Manganese 0.31 

Aluminium 6.50 

Copper 1.05 

Sulphur 0.00 

Phosphorus 0.13 
















The copper in this alloy was present by accident, the alloy 
regularly made containing none, but the rest of the analysis 
gives a correct idea of the constitution of the alloy. Prof. 
Mabery gives several analyses of Cowles' ferro-aluminium : — * 

Iron 85.17 

Aluminium 8.02 

Silicon 2.36 


The slags formed in the furnace in producing this alloy were 
analyzed as follows : — 

Silica 0.78 4.10 

Alumina (insoluble) 0.20 — 

Lime 28.50 14.00 

Iron 1.50 29.16 

Alumina (soluble) -(- aluminium 38.00 48.70 

Sulphur 0.50 — 

Graphite 5.00 2.60 

Combined carbon 0.90 0.48 

The slags formed when producing bronze vary in composition, 
"American Chemical Journal, 1887, p. 11. 


and are usually crystalline, with a shining, vitreous lustre. 
Their analysis shows — 

Alumina (insoluble) 55.30 66.84 — 

Alumina (soluble) + aluminium 21.80 14.20 — 

Lime 3.70 1.44 6.77 

Copper — 3.32 i.oo 

Carbon 0.65 

The lime present probably existed as calcium aluminate. These 
slags contained only a small amount of aluminium, rarely any 
iron, and were usually free from silica. 

The same chemist analyzed a peculiar product sometimes 
formed in the furnace when smelting for bronze, in the shape 
of crystalline masses, steel-gray to bright yellow in color, semi- 
transparent and with a resinous lustre. These all contained 
aluminium, copper, silicon and calcium in various proportions, 
and when exposed to the air fell to powder. Analyses gave 

Copper 26.70 35-0° 20.00 — 

Aluminium 66.20 53-30 74.32 15.23 

Silicon 5.00 12.30 2.86 20.55 

Calcium 2.00 0.20 2.86 ^ 

Tin _ _ _ 49.26 

99.90 100.80 100.04 85.04 

The latter product was formed in smelting for aluminium-tin. 

Prof. Mabery also found that the soot collecting at the ori- 
fices on top of the furnace contained 10 to 12 per cent, of alu- 
minium ; also that when alumina and carbon alone were heated 
and silica was present, the aluminium formed dissolved up to 
10 per cent, of silicon, which, on dissolving the aluminium in 
hydrochloric acid, was left as crystalline or graphitic silicon. 

Reactions in Cowles' process. — The inventors, themselves, 
claim " reduction in a furnace heated by electricity in presence 
of carbon and a metal." In their first pamphlet they say that 
" the Cowles process accomplishes the reduction of alumina by 
carbon and heat." Professor Mabery and Dr. Hunt, already 


quoted, and Dr. Kosman * look at the process in no other light 
than that the electric current is utilized simply by its conver- 
sion into heat by the resistance offered, and that pure elec- 
trolysis is either absent or occurs to so small an extent as to be 
inappreciable. Indeed, if we consider the arrangement of the 
parts in the Cowles furnace we see every effort made to oppose 
a uniform, high resistance to the passage of the current and so 
convert its energy into heat, and an entire absence of any of 
the usual arrangements for electrolysis. For instance, elec- 
trolysis requires a fluid bath in circulation, so that each ele- 
ment of the electrolyte may be continuously liberated at one of 
the poles and the presence of any foreign material, as bits of 
carbon, between the poles is to be avoided if possible, since 
they short-circuit the current and hinder electrolysis proper. 
I think the arrangement of the furnace shows no attempt to 
fulfil any of the usual conditions for electrolysis, but is one of 
the best arrangements for converting the energy of the current 
entirely into heat. Dr. Hampe, however, in spite of these 
evident facts, draws the conclusion that because he was unable 
to reduce alumina by carbon in presence of copper at the tem- 
perature of a Deville lime-furnace, that it was therefore to be 
assumed that even the somewhat higher temperature of the 
electric furnace alone would be insufficient to accomplish the 
desired reaction, and hence the effect of the electric arc must 
be not only electro-thermic in supplying heat, but afterwards 
electrolytic, in decomposing the fused alumina. 

If we figure out the useful effect of the current, i. e., the pro- 
portion of its energy utilized for the purpose of reducing alumina, 
we find a low figure ; but it is well to note that although the 
power required is one of the main features of this way of reduction, 
yet this item is so cheap at the firm's works that it becomes a 
secondary consideration in the economy of the process. A 300 
horse-power current is equivalent to an expenditure of 

300 7 — ?__ jg J 000 calories of heat per hour. Theoretically, 


* Stahl und Eisen, Jan. I 


this amount of heat would produce i^i^ = 261^ kilos or 58 

pounds of aluminium. However, about 7 pounds are obtained 
in an hour's working, which would show a useful effect of 12 
per cent. This should even be diminished, since no account 
has been taken of the combustion of carbon in the furnace to 
carbonic oxide. The remainder of the heat account, probably 
90 per cent, of the whole, is partly accounted for by the heat 
contained in the gases escaping and the materials withdrawn 
from the furnace (of which no resonable estimate can be made, 
since the question of temperatures is so uncertain) and the 
large remainder must be put down as lost by radiation and 
conduction. As before remarked, water power is obtained by 
this company very cheaply, and even this large loss does not 
make much show in the cost of the alloy, yet the figures show 
that a much larger useful effect should be possible, and it is not 
at all improbable that the prospect of getting double or triple 
the present output from the same plant is at present inciting 
the managers to fresh exertions in utilizing the power to better 

Mr. H. T. Dagger's paper* on the Cowles process in Eng- 
land, states that the product at their Milton works is i lb. of 
aluminium to 18 electric horse-power per hour, which would 
show that the dissociation of the alumina represented nearly 30 
per cent, of the energy of the current ; but the data given in the 
body of this gentleman's paper (p. 303 ) do not seem to indi- 
cate so large a return as is stated above. Mr. Dagger, more- 
over, maintains the purely electro-thermic action of the current, 
denying that any electrolysis takes place at all, citing an ex- 
periment with the alternating current in one of these furnaces, 
which produced as much aluminium per horse-power as did the 
direct current. In such an experiment electrolysis must neces- 
sarily be absent. 

In the discussion of Heroult's alloy process it will be shown 
that in both it and Cowles' process the largest part of the re- 

* British Association for Advancement of Science, Newcastle, 1889. 


duction must necessarily be performed by chemical and not by 
electrolytic action. I do not introduce this discussion here, 
since the two processes resemble each other so closely in the 
reaction involved that they can best be considered together. 

Since 1892 no aluminium has been made in England by this 
process, as the low price of pure aluminium killed the market 
for ready-made aluminium alloys. I cannot say exactly to 
what extent the Lockport works is running at the present time, 
but it is probable that the death of one of the Cowles brothers 
in 1893 and the sharp competition from pure aluminium have 
seriously hampered the business of the American company. 
At least, their products are not frequently seen on the market. 

It may be proper to interject here mention of an unfortunate 
occurrence in the history of the Cowles Company. They sold 
only aluminium alloys until January, 1891, when they began 
to advertise and sell pure aluminium made, it was said, by a 
new process. The Pittsburgh Reduction Company working 
the Hall process were then selling pure aluminium as low as 
$1.50 per pound, while the Cowles Company commenced sell- 
ing as cheap as $1.00. Inside of two months the Pittsburgh 
Reduction Company instituted suit for infringement of the 
Hall patents, asking the courts meanwhile for a preliminary 
injunction. Judge Ricks, of Ohio, denied a complete injunc- 
tion, but restrained the defendants from increasing the output 
of their plant during the trial of the suit, or from selling alu- 
minium below a price to be named by the complainants. The 
latter named $1.50 per pound, and this fixed the market price 
for some months. Several metallurgical experts of wide repu- 
tation were engaged by both sides, and very voluminous testi- 
mony was taken. While the suit was pending, the price of 
aluminium was reduced by the European makers to 5 marks a 
kilo (56 cents a pound), at which price important quantities 
began to be imported, and the domestic makers could sell very 
Httle at $1.50. In consequence, the Pittsburgh Reduction 
Company notified the court that it would sell as low as $0.50 
per pound. At this price, which was very nearly the cost of 


production, aluminium continued until the suit was finally de- 
cided, in February, 1893, in favor of the Hall process. Judges 
Taft and Ricks decided that the Cowles Company were using 
the Hall process when they electrolyzed a bath of molten 
cryolite in which alumina was dissolved, and they were ordered 
to stop the infringement and to pay the damages which it 
might be estimated that their infringement had caused the 
complainants' business. 

Menges' Patent. 

* This inventor proposes to produce aluminium or aluminium 
bronze by mixing aluminous material with suitable conducting 
material, such as coal, and a cohesive material, then pressing 
into cylinders and baking hard. These strong, compact bars 
conduct electricity, and are to be used like the carbon elec- 
trodes of electric lamps in a suitably inclosed space. 

Farmer's Patent. 

M. G. Farmer f mixes aluminous material with molasses or 
pitch, making a paste which is moulded into sticks, burned, 
and used as electrodes, inclosed in a furnace. Aluminium is 
produced by the arc, and drops into a crucible placed imme- 
diately beneath. 

Several other persons have patented exactly the same 
method of procedure as that indicated by Menges and Farmer, 
with the modification in some cases of mixing iron or copper 
turnings with the aluminous material, and having the melted 
material drop into a furnace kept hot by burning ordinary fuel. 
Such a method of reducing refractory ores was even suggested 
as far back as 1853. None of such processes are likely to give 
results sufficiently economical for commercial use. 

Kleiner's Process (1886). 
This was devised by Dr. Ed. Kleiner of Zurich, Switzerland, 

* German Patent, 40354 (1887). 

t English Patent, 10815, Aug. 6, 1887. 


and was patented in most of the European States. The Eng- 
lish patent is dated 1886.* The first attempts to operate it 
were at the Rhine Falls, Schaffhausen, and were promising 
enough to induce Messrs. J. G. Nethers, Sons & Co., pro- 
prietors of an iron works there, to try to obtain water rights for 
1500 horse power, announcing that a company (the Kleiner 
Gesellschaft) with a capital of 12,000,000 francs, was prepared 
to undertake the enterprise and build large works. The pro- 
position is said to have met with strong opposition from the 
hotel-keepers and those interested in the falls as an attraction 
for tourists, and the government declined the grant, consider- 
ing that the picturesqueness of the falls would be seriously 
afifected. This is the reason given by those interested in the 
process for its not being carried out in Switzerland, it being 
then determined to start a works in some part of England 
where cheap coal could be obtained, and test the process on a 
large scale. A small experimental plant was then set up in the 
early part of 1887 on Farrington Road, London, where it was 
inspected by many scientific men, among them Dr. John Hop- 
kinson, F. R. S., who reported on the quantitative results 
obtained ; a description of the process as here operated was 
also written up for " Engineering." With the co-operation of 
Major Ricarde-Seaver a larger plant was put up at Hope Mills, 
Tydesley, in Lancashire, where the process was inspected and 
reported on by Dr. George Gore, the electrician. After his 
report we learn that the patents were acquired by the Alumin- 
ium Syndicate, Limited, of London, a combination of capitalists 
among whom are said to be the Rothschilds. 

The aluminium compound used is commercial cryolite. It 
is stated that the native mineral from Greenland contains on an 
average, according to Dr. Kleiner's analysis, 96 per cent, of 
pure cryolite, the remainder being moisture, silica, oxides of 
iron and manganese. As pure cryolite contains 13 per cent, 
of aluminium, the native mineral will contain 12^ per cent., 

♦English Patents, 8531, June 29, 1886, and 15322, Nov. 24, 1886. 



all of which Dr. Kleiner claims to be able to extract. It is 
further remarked that as soon as sufificient demand arises, an 
artificial cryolite can be made at much less cost than that of the 
native mineral, which now sells at ;^i8 to ;^20 a ton. The 
rationale of the process consists in applying the electric current 
in such a way that a small quantity of it generates heat and 
keeps the electrolyte in fusion, while the larger quantity acts 
electrolytically. Dry, powdered cryolite is packed around and 
between carbon electrodes in a bauxite-lined iron crucible ; on 

Fig. 35. 

passing a current of high tension (80 to 100 volts) through the 
electrodes, the cryolite is quickly fused by the heat of the arc 
and becomes a conductor. As soon as the electrolyte is in 
good fusion the tension is lowered to 50 volts, the quantity 
being about 150 amperes, the arc ceases and the decomposi- 
tion proceeds regularly for two or three hours until the bath is 



nearly exhausted. The evolved fluorine is said to attack the 
bauxite, and by thus supplying aluminium to the bath extends 
the time of an operation. In the first patent the negative car- 
bon was inserted through the bottom of the melting cavity, the 
positive dipping into the bath from above, (Fig. 35) but it was 
found that while the ends of the positive carbon immersed in 
the cryolite were unattacked, the part immediately over the 
bath was rapidly corroded. In the second patent, therefore, 
the positive electrode was circular and entirely immersed in 
the cryolite, connection being made by ears which projected 
through the side of the vessel. (Fig. 36.) As the carbons 

Fig. 36. 

are thus fixed, the preliminary fusion is accomplished by a 
movable carbon rod suspended from above, passing through 
the circular anode and used only for this purpose. The bath 
being well fused and the current flowing freely between the 
fixed carbons, the rod is withdrawn. The carbons are said to 
be thus perfectly protected from corrosion, and able to serve 
almost indefinitely. The melting pots finally used were or- 
dinary black-lead crucibles, which are not usually injured at all, 
since the fused part of the cryolite does not touch them, and 
they last as many as 300 fusions. After the operation, the 
carbons are lifted out of the bath and the contents cooled. 


When solid, the crucibles are inverted and the contents fall 
out. This residue is broken to coarse powder, the nodules of 
aluminium picked out, melted in a crucible and cast into bars. 
The coarse powder is then ground to fine dust. This powder 
is more or less alkaline and contains a greater or less excess of 
fluoride of sodium in proportion to the amount of aluminium 
which has been taken out. If only a small proportion of the 
metal has been extracted and the powder contains only a small 
excess of sodium fluoride, it is used again without any pre- 
paration in charging the crucibles ; but if as much as 5 or 6 
per cent, of aluminium has been removed and the powder, 
therefore, contains a large excess of sodium fluoride, it is 
washed with water for a long time to remove that salt, which 
slowly dissolves. The solution is reserved, while the powder 
remaining is unchanged cryolite, and is used over. Dr. Gore 
states that if the powder, electrodes and crucible are perfectly 
dry, there is no escape of gas or vapor during the process ; 
but if moisture is present, a small amount only of fumes of 
hydrofluoric acid appear, and that there is no escape of fluorine 
gas at any time. Dr. Kleiner hopes to soon dispense with the 
interruption of the process, washing, etc., by regenerating 
cryoHte in the crucible itself and so making the process con- 
tinuous. One of the great advantages claimed is that the 
aluminium is obtained in nodules, and not in fine powder; if it 
was, it could not all be collected because it is so light, some of 
it would float upon the water during the washing process and 
be lost, and even when collected it could not be dried and 
melted without considerable loss. 

It has been found impossible in practice to obtain all the 
aluminium from a given quantity of cryolite in less than two 
fusions, for the sodium fluoride collecting in the bath hinders 
the production of the metal. The proportion extracted by a 
single fusion depends upon its duration. In the operations at 
Tydesley, a fusion lasting 24 hours separated only 2^ per 
cent, of aluminium, whereas the cryolite contained 12^ per 
cent. At this rate, to extract the whole in two operations would 


require two fusions of 6o hours each. As to the output, on an 
average a current of 38 electric horse-power deposited 150 
grammes of aluminium per hour, being a little over 3 grammes 
per horse-power. Since a current of 50 volts and 150 amperes, 
such as was stated above as the current in each pot, is equal to 
50x^50 ^^ j^ electric H. P., it is probable that the 38 H. P. cur- 
rent mentioned must have been used for four crucibles. Now, 
the output of four crucibles, each with a current of 150 amperes, 
should have been 0.00009135X150X4=0.0548 gramme per 
second, or 197.3 grammes per hour; the difference between 
this and the amount actually obtained, or 47.3 grammes, is the 
amount of aluminium which was produced and then afterwards 
lost either as fine shot-metal or powder, or dissolved again by 
corroding elements in the bath. To calculate how the output 
of 3 grammes per electric H. P. per hour compares with the 
quantity of metal which this amount of energy should be able 
to produce, we can assume that since it takes at most 4 volts to 
decompose the aluminium fluoride, the four crucibles would 
consume 16 volts in decomposition, being sixteen-fiftieths of 
the total voltage employed. Since the amperes only produced 
150 grammes out of 197.3 theoretically possible, the efhciency 
over all is only 

16 150 1 

As to the purity of the metal obtained, the process is met at 
the outset by the silica and iron oxide in the cryolite, which are 
probably all reduced with the first few grammes of aluminium 
thrown down. This can possibly be remedied by using a purer 
artificial cryolite ; the impurities cannot generally be separated 
from the natural mineral. Then there are impurities of a simi- 
lar nature coming from the carbons used, and which are gene- 
rally present if especial pains are not taken to get very pure ma- 
terials for making them. Dr. Kleiner's early attempts produced 
metal of 85 to 95 per cent, purity, but he stated in 1889 that it 
was uniformly 95 to 98 per cent., and being put on the market in 


competition with other commercial brands. It appears, how- 
ever, from a consideration of the preceding data, that the pro- 
cess could not produce aluminium much cheaper than 12 shil- 
lings a pound, and it was therefore abandoned in 1890. It may 
be observed that the mechanical arrangements for carrying on 
this process, as above outlined, could not stand comparison with 
the present forms of apparatus in use by the Hall and Heroult 

Lossier's Method. 

* This is a device for decomposing the natural silicates by 
electricity and obtaining their aluminium. The bath is com- 
posed of pure aluminium fluoride or of a mixture of this salt 
and an alkaline chloride, and is kept molten in a round bot- 
tomed crucible placed in a furnace. The electrodes are of dense 
carbon and are separated in the crucible by a partition reach- 
ing beneath the surface of the bath. The positive electrode is 
furnished with a jacket or thick coating of some aluminium sili- 
cate plastered on moist and well dried before use. When the 
current is passed, the aluminium fluoride yields up its fluorine at 
this pole and its aluminium at the other. The fluorine com- 
bines with the aluminium silicate, forming on the one hand alu- 
minium fluoride, which regenerates the bath, on the other silicon 
fluoride and carbonic oxide, which escape as gases. The metal 
liberated at the negative pole is lighter than the fused bath, and 
therefore rises to the surface. 

M. Grabau cites as one of the recommendations of aluminium 
fluoride for use in his process (p. 295), that it is quite infusible, 
so it would appear that Lossier has made a mistake in suppos- 
ing that it could be melted alone in a crucible. It would, how- 
ever, make a very fusible bath when the alkali chloride was 
added. It is probable that the carrying out of this method 
would develop great trouble from the attacking of the crucible 
by the very corrosive bath, the disintegration of the carbons, 
which would cause much trouble at the negative pole especially, 

* German Patent (D. R. P.), No. 31089. 


and the oxidation of the fluid aluminium on the surface of the 
bath. The process has never been attempted on a large scale, 
and it is very unlikely that it ever will be. 

OmhoKs Furnace. 

I. Omholt and the firm Bottiger and Seidler, of Gossnitz, have 
patented the following apparatus for the continuous electrolysis 
of aluminium chloride:* — 

The bed of a reverberatory furnace is divided by transverse 
partitions into two compartments, in each of which are two re- 
torts semi-circular in section, lying side by side horizontally 
across the furnace, with the circular part up. They are sup- 
ported on refractory pillars so that their open side is a small 
distance above the floor of the furnace. The aluminium com- 
pound being melted on the hearth, it stands to the same depth 
in both retorts, and if the electrodes are passed through the 
bottom of the hearth they may remain entirely submerged in 
molten salt and each under its own retort cover. The metal 
therefore collects in a liquid state under one retort and the 
chlorine under the other, both being preserved from contact or 
mixture with the furnace gases by the lock of molten salt. The 
chlorine can thus be led away by a pipe, and utilized, while the 
aluminium collects without loss, and is removed at convenient 

The great cost of making aluminium chloride will probably 
prevent it ever being used again in the metallurgy of alumin- 
ium, but the form of furnace described above may possibly be 
utilized in operating the decomposition of some other aluminium 
salt, such as the double sulphide with soda. 

Minefs Process (1887). 
This consists, according to the patent specification, f in the 
electrolysis of a mixture of sodium chloride with aluminium 
fluoride or with the separate or double fluorides of aluminium 

* German Patent (D. R. P.), No. 34728. 
t English Patent, No. 10057, July r8, 1887. 



and sodium, melted in a non-metallic crucible or in a metallic 
one inclosed in a thin refractory jacket to avoid filtration, the 
aluminium fluoride decomposed being regenerated by causing 
the fluorine vapors evolved to act on bauxite or alumina 
placed somewhere about the anode. The details of the appar- 
atus and bath are as follows : — 

Disposition of the Apparatus. — The pots or crucibles used 
may be of refractory earth, plumbago, or of metal, and in cases 
where an alloy is required the crucible itself serves as an elec- 
trode. None of these, however, resist the corrosive power of 
the electrolyte and would under ordinary conditions be quickly 
destroyed. To overcome this difficulty two special devices are 
employed. When alloys are to be made directly, the pot is 
cast of the metal with which the aluminium is to be combined. 
It is shaped with a sloping bottom and provided with a tap 
hole. The pot is encased in thin brickwork and is then made 
the negative electrode, the positive being two carbon rods 
dipped into the bath. As soon as the current is passed alu- 
minium is deposited on the walls of the pot, forming a rich 
alloy with the metal of which the pot is made (iron or copper). 
When this coating becomes sufficiently rich in aluminium, the 
heat of the bath melts it and it trickles down and collects at the 
bottom. After a certain time, the alloy can be tapped out 
regularly at intervals without interrupting the electrolysis. The 
metal thus obtained is principally aluminium containing a few 
per cent, of the metal of the pot, which is of no consequence, 
since the end to be finally attained is the production of an alloy 
with a smaller quantity of aluminium. When pure aluminium is 
to be obtained, an ingenious device is used to protect the metal 
from contamination by the metal of the pot. Carbon plates, A 
and C (Fig. 37), serve as anode and cathode, the cathode stand- 
ing upright in a small crucible placed upon a plate resting on 
the bottom of the pot. This crucible and plate are made from 
'carbon blocks or from fused alumina or fluorspar moulded into 
the shape desired. As the metal is set free it trickles down 
the cathode and is caught in the crucible or cup, thus being 



prevented from spreading out over the bottom of the pot. To 
prevent the bath from corroding the pot, the wire R is passed 
from the latter to the negative pole of the circuit. The pot is 
thus made part of the negative electrode, but it is not intended 
that much of the current should pass through it, so a resistance 
coil is interposed between it and the battery or dynamo, so that 

Fig. 37. 

the derived current passing through the sides of the pot is only 
5 to 10 per cent, of the whole current. The effect of this is 
that a small amount of aluminium is deposited on and alloys 
with the sides of the vessel, which protects the latter from cor- 
rosion and is only feebly acted upon by the bath. The metal 
deposited in the crucible is thus kept nearly pure, while a small 
amount of alloy falls to the bottom of the pot and is poured 
out after the crucible has been removed. When it is wished to 
obtain the purest aluminium, the intensity of the derived cur- 
rent passing through the pot is increased by removing part of 
the resistance interposed between it and the negative wire, thus 
also decreasing the intensity of the principal current. The 


nature of the electrodes proper may be varied. For producing 
pure aluminium the anode is carbon, the cathode carbon, and 
the pot either of copper or iron ; for producing copper alloys 
the anode may be either carbon or bright copper, and the 
cathode (pot) of carbon or copper; for producing iron alloys 
the anode may be either carbon or iron, while the vessel used 
as cathode is either of cast-iron or plumbago. 

Composition and properties of the baths. — Minet has made 
more careful electrical measurements on these fused baths than 
any other metallurgist, a point of particular importance in his 
experiments being the accurate measurement of the tempera- 
tures by means of a Le Chatellier electric pyrometer. On this 
account his result^re a valuable contribution to the knowledge 
of the present electrolytic processes, and are therefore worth 
repeating in extenso. Minet first experimented with a bath of 

Aluminium-sodium chloride 40 parts. 

Sodium chloride 60 " 

He says of the working of this mixture that it gives rise to 
abundant vapors, which are very disagreeable. In consequence, 
the bath becomes quickly impoverished in aluminium chloride, 
it becomes pasty, and can only be liquefied by raising the tem- 
perature almost to the melting point of sodium chloride. It is 
difficult under these conditions to produce regular and continu- 
ous electrolysis. 

Minet obtained much better results with aluminium fluoride. 
He mixed cryolite with common salt in the following propor- 
tions : 

Cryolite 37-5 parts. 

Sodium chloride 62.5 " 

giving a mixture whose formula would be AlFgSNaF + 6NaCl. 
This mixture has the following properties : 

Melting point 675° C. 

Vapors given off at 1056° C. 

Specific gravity at 829° C 1.76 

Specific resistance at 870° C 0.32 ohms. 


The last datum means that a column of the electrolyte i 
centimetre square and i centimetre long would give 0.32 ohms 
resistance to the passage of the current. If the section of the 
electrolyte through which the current passed averaged 750 
square centimetres, and the electrodes averaged 6 centimetres 
apart, the total resistance of the bath would be 

0.32 X ^0.0026 ohms: 

and if the current was 1 500 amperes, the voltage absorbed by 
the electrical conduction-resistance of the bath would be 

0.0026 X 1500 = 3.9 volts. 

The specific resistance was found to decrease about 0.07 per 
cent, for every degree rise of temperature above 870°, or 7 per 
cent, for every lOO degrees. 

If this bath were simply electrolyzed, fluorine would be dis- 
engaged, escaping into the air as carbon fluoride, CF4, while 
the bath would become poorer in aluminium fluoride. If cryo- 
lite alone were added during the process, sodium fluoride would 
accumulate in the bath, and after a certain time sodium would 
be set free. If the bath is renewed by adding aluminium fluor- 
ide, this inconvenience is obviated, but there still remains the 
question of suppressing the noxious fluoride vapors set free. If 
the bath is renewed by placing alumina in the state of fine pow- 
der in the bath, dropping it especially around the anode, the 
fluorine set free may reform aluminium fluoride by attacking 
the alumina, setting free oxygen, which combines with the car- 
bon anode to form carbonic oxide. If the fluorine were not 
completely absorbed by the alumina, the amount escaping would 
have to be made up for by adding aluminium fluoride to the 
bath. The bath should therefore be fed by both alumina and 
aluminium fluoride. 

The above contains Minet's views of what takes place when 
alumina is added to the fluoride bath. He does not admit that 
any alumina can be really dissolved in the bath and be electro- 


lyzed as alumina (the view taken by Hall). Minet, in fact, 
maintains that when alumina is added to molten cryolite it 
sinks through it and remains undissolved, as, for instance, very 
fine marble dust would in water. Such, however, is not the 
experience of other investigators. According to numerous 
experiments by Hall and others using the present electrolytic 
processes, as well as some made by the writer, finely-divided 
alumina appears to dissolve in molten cryolite exactly as sugar 
does in water, and there is even a saturation point, at about 30 
per cent., above which no more dissolves. The bath containing 
alumina forms to all appearances a homogeneous liquid. 
Driven from their first position, the adherents to Minet's views, 
having allowed that a homogeneous liquid bath does result, as- 
sert that the alumina is only mechanically suspended, like 
clay in muddy water, and that in this condition it is carried to 
the anode and unites with the fluorine set free. Such a view, 
however, is also erroneous, because regeneration in such a man- 
ner could not be anything like complete, and considerable 
amounts of fluorine would escape ; yet in the Hall process, 
when the operation is properly conducted, not a trace of fluoride 
gas escapes. Regeneration by mechanically-suspended alu- 
mina could not possibly be so complete. 

We shall continue describing Minet's experiments, asking that 
it be remembered, however, that we regard as erroneous his 
view that aluminium fluoride is the substance immediately de- 
composed by the current when alumina is present. 

Taking the bath already described, with electrodes of such 
size and distance apart that the conduction-resistance at 900° 
was 0.0044 ohms, the following results were obtained : — 


Electromotive force 


for decomposition. 

of the bath. 


2.4 volts 

.0044 ohms 


2.34 " 

.0033 " 


2.17 " 

.0025 " 

If V' = the voltage required for decomposition, 
V° = the voltage required for conduction, 
V = the total voltage absorbed by the bath, 
A =; the amperage of the current. 

then [V = VI -(- AV".] 



We can thus calculate the voltage required to send through 
this bath any number of amperes at a given temperature. 
Thus, for lOO or looo amperes the voltage absorbed per bath 
would be as follows : 

Temperature. Amperes. Voltage for 



1 100" 













iltage for 


Per cent, of total 



voltage absorbed 
in conduction re- 



















It thus appears that, using a bath of a given size, the voltage 
required to send lOOO amperes through it is only 2 to 2.5 times 
that required for 100 amperes. Since the power required is 
just in the same ratio, it follows that it is economical to send as 
large a current as practicable through a bath of a given size. 
The last column shows us the percentage of the energy of the 
current which will be converted into heat. It follows from 
these figures that the heating effect of the current increases 
very rapidly with its quantity, so that a given number of 
amperes will supply enough heat to keep the bath molten with- 
out exterior heat; more than this number would, in fact, raise 
the temperature of the bath too high, and this limits the size of 
the current which it is practicable to send through a bath of a 
given size. 

Minet also found that when such a bath contained salts of 
iron and silicon, which are weaker compounds than aluminium 
or sodium salts, a current of carefully regulated low voltage, and 
consequently of low density, would first separate these com- 
pletely. An experiment made with a bath in which the anode 
surface was 500 square centimetres, and gradually increasing 
the voltage, gave the following successive results : 


Total Amperes Amperes per Electro-motive Nature of the 

voltage passing. sq. c. m. of force of decom- metal deposited. 



position (volts). 









Iron ( traces of silicon) . 










Ferro-silicon (traces of aluminium) 





Silicon-aluminium (traces of iron) . 





Aluminium (traces of silicon) . 



1. 00 


Aluminium (traces of sodium). 

The conduction-resistance of the bath remained very nearly 
constant throughout. 

In a large number of experiments, the amount of metal actu- 
ally obtained was 50 to 80 per cent, of the amount which 
the amperage of the current could theoretically produce, on an 
average about 60 per cent. As for the power required, an 
experiment lasting 24 hours, at a temperature of 1 160°, voltage 
5.75, and amperes 1500, showed 21.5 grammes of aluminium 
deposited per hour per electric horse-power used. Another 
similar experiment, but with the electrodes closer together, the 
voltage consequently less, and at a lower temperature, gave 31.9 
grammes per horse-power-hour, which would be 0.765 kilos 
(1.685 pounds) per electric horse-power per day. As the en- 
ergy of one horse-power-day is equivalent to the energy of oxi- 
dation of 2.1 kilos (4.63 pounds) of aluminium, the useful efifect 
over all is 36.5 per cent. 

Quality of metal. — When working for pure aluminium, about 
three-fourths of the metal produced is taken from the crucible 
in which the cathode stands, and is 98 to 99 per cent, pure ; 
the other one-fourth has been deposited on the sides of the 
cast-iron pot, and contains 10 to 20 per cent, of iron. It is 
poured out and used for making ferro-aluminium. 

Installations of the process. — Minet conducted his work at the 
expense of Messrs. Bernard Bros. The first experiments were 
made on the road Moulin Joli, in Paris, from March, 1887, to 
March, 1888. This plant consisted of a 6 horse-power engine 
and a Gramme dynamo capable of giving a current of 250 
amperes at 12 volts tension. In the year named, this plant pro- 


duced about 500 kilogrammes of pure aluminium and 1500 
kilogrammes of ferro-aluminium or aluminium bronze. 

The second installation was put up at Creil (Oise), com- 
menced operations in April, 1888, and continued working until 
October, 1891. This plant had a 40 horse-power engine, and 
an Edison dynamo giving a current of 1200 amperes and 27.5 
volts. Regularly it was run at only 16 volts, operating three 
baths in tension, and at looo amperes. There were made here 
daily about 10 kilos of pure aluminium and 5 to 6 kilos of alu- 
minium in alloys. 

At the first location, Minet was enabled to study experiment- 
ally the purely scientific part of his processes ; the plant at 
Creil enabled him to solve the practical details of the process. 
In 1889, the Messrs. Bernard Bros, determined to work the 
process on a large, industrial scale, and cast around for a loca- 
tion having abundant water-power. It was decided to locate at 
Saint Michel in Savoy. The stream utihzed is the Valoirette, 
giving a volume of 3.5 cubic metres per second with a fall of 
133 metres. This gives theoretically 6000 horse-power, and 
should produce 4000 electric horse-power. 

At this place were put up a turbine of 300 horse-power by 
Bouvier of Grenoble, and a dynamo of 275 horse-power by 
Hillairet-Huguet of Paris. The latter gives a current of 4000 
amperes at 50 volts tension, aiid produces regularly about 150 
kilos (330 pounds) of aluminium. It is stated that a 600 horse- 
power dynamo, actuated by a lOOO horse-power turbine, was 
started during 1894. Latest advices state that this estabUsh- 
ment has been acquired by a French company formed to work 
under the Hall patents. 

Feldman's Method (1887). 

A. Feldman, of Linden, Hannover, patented the following 
electrolytic process : * — 

A double fluoride of aluminium and an alkaline earth metal, 

* English Patent, No. 12575, Sept. 16, 1887. 


mixed with an excess of a chloride of the latter group, is either 
electrolyzed or reduced by sodium. The proportions of these 
substances to be used are such as take place in the following 
reactions : — 

1. (AUF6+2SrF2) + 6SrCI.,=2Al+5SrF2+3SrCl2+6Cl. 

2. (AljF6+2SrF2)+6SrCl2+6Na=2AH-sSrF2+3SrCl2+6NaCl. 

The three equivalents of strontium chloride are found in prac- 
tice to be most suitable. Potassium chloride may also be 
added to increase the fluidity, but in this case the strontium 
chloride must be in still greater excess. 

Even if the above reactions and transpositions do take place, 
the use of so much costly strontium salts would appear to ren- 
der the process uneconomical. 

Warren's Experiments (1887). 

Mr. H. Warren, of the Everton Research Laboratory, has 
outlined the following methods or suggestions, some of which 
had already been carried out, and probably others have since 
given useful ideas to workers in this line. The principle can 
hardly be called new, since suggestions almost identical with 
Mr. Warren's were made previously to his, but the latter's re- 
sults are the first recorded in this particular direction:* "This 
method of preparing alloys differs only slightly from the man- 
ner in which amalgams of different metals are prepared, substi- 
tuting for mercury the metals iron, copper, or zinc made liquid 
by heat. These metals are melted, connected with the negative 
pole of a battery, and the positive pole immersed in a bath of 
molten salt floating on top of the melted metal. The apparatus 
used is a deep, conical crucible, through the bottom of which 
is inserted a graphite rod, projecting about one inch within, the 
part outside being protected by an iron tube coated with borax. 
As an example of the method, to prepare silicon bronze, copper 
is melted in the crucible, a bath of potassium silico-fluoride is 

* Chemical News, Oct. 7, 1887. 


fused on top to a depth of about two inches. A thick platinum 
wire dips into this salt, and on passing the electric current an 
instantaneous action is seen, dense white vapors are evolved, 
and all the silicon, as it is produced, unites with the copper, 
forming a brittle alloy. Cryolite may be decomposed in like 
manner if melted over zinc, forming an alloy of zinc and alu- 
minium, from which the zinc can be distilled, leaving pure 

Mr. Warren does not affirm that he has actually performed 
the decomposition of cryolite in the way recommended, but 
states that it may be done ; from which we would infer that he 
simply supposed it could. A well-recorded experiment, then, 
is needed to establish the truth of this statement. Neither does 
he propose to make aluminium bronze in this way; it may be 
that it was attempted and did not succeed, for Hampe states 
that an experiment thus conducted did not furnish him alu- 
minium bronze (p. 368). 

Zdziarski' s Patent. 

A. Zdziarski,* of Brest-Litowsk, Russia, appears to have 
patented the above principle in 1884, for in his patent he states 
that the metal to be alloyed with aluminium is melted in a 
crucible, covered with a fusible compound of aluminium for a 
flux (alumina and potassium carbonate may be used) and made 
the negative pole of an electric current, the positive pole being 
a carbon rod dipping in the flux. 

Grabau's Apparatus. 

LudwigGrabaUjf of Hannover, Germany, proposes to electro- 
lyze a molten bath of cryolite mixed with sodium chloride. The 
features of the apparatus used are an iron pot, in which the 
bath is melted, and water-cooled cylinders surrounding both 
electrodes, the jacket surrounding the negative one having a 

* English Patent 3090, Feb. 11, 1884. 
t German Patent (D. R. P.), No. 45012. 


bottom, the other not. The object of these cylinders is, at the 
positive electrode, to keep the liberated fluorine from attacking 
the iron pot and so contaminating the bath, at the other pole 
the liberated aluminium is kept from dropping to the bottom 
of the pot, where it might take up iron, and can be removed 
from the bath by simply lifting out the water-cooled cylinders 
and carbon electrode. Mr. Grabau states that he has abandoned 
this process because the inseparable impurities in the cryolite 
produced impurities in the metal; it may be that with the pure 
artificial cryolite, which he makes by his other processes (see 
p. 171), this electrolytic process may again be taken up. (See 
also p. 294.) 

Rogers' Process (1887). 

In July, 1887, the American Aluminium Company, of Mil- 
waukee, was incorporated, with a capital stock of $1,000,000, 
for the purpose of extracting aluminium by methods devised 
by Prof. A. J. Rogers, a professor of chemistry in that city. 
This gentleman had been working at the subject for three or 
four years previous to that time, but it has not been until re- 
cently that patents have been applied for, and they are still 

The principle made use of has already been suggested in con- 
nection with the production of sodium (p. 219). It is briefly, 
that if molten sodium chloride is electrolyzed, using a molten 
lead cathode, a lead-sodium alloy is produced. This alloy is 
capable of reacting on molten cryolite, setting free aluminium, 
which does not combine with the lead remaining because of its 
weak affinity for that metal. If, then, cryolite is placed in the 
bath with the sodium chloride, the two reactions take place at 
once, and aluminium is produced. In the early part of 1888, 
the company erected a small experimental plant, with a ten- 
horse-power engine, with which the following experiments, 
among many others, were made: — 

I. *A current of 60 to 80 amperes was passed for several 

* Proceedings of the Wisconsin Nat. His. Soc, April, 1889. 


hours through a bath of cryolite melted in a crucible lined with 
alumina, and using carbon rods 2^ inches in diameter as 
electrodes, one dipping into the bath from above, the other 
passing through the bottom of the crucible into the bath. 
Only I or 2 grammes of aluminium were obtained, showing 
that the separated metal was almost all redissolved or reunited 
with fluorine. With the temperature very high, it was found 
that sodium passed away from the bath without reducing the 

2. A current averaging 54 amperes and 10 volts was passed 
for five and a half hours through a mixture of I part cryolite 
and 5 parts sodium chloride placed in a crucible with 3 70 
grammes of molten lead in the bottom as the cathode. After 
the experiment, 25 grammes of aluminium were found in 
globules on top of the lead-sodium alloy. This latter alloy 
contained some aluminium. The globules were about as pure 
as ordinary commercial aluminium, and contained no lead or 
sodium. From another experiment it was determined that the 
lead-sodium alloy must first acquire a certain richness in 
sodium before it will part with any of that metal to perform 
the reduction of the cryolite. It was also found that a certain 
temperature was necessary in order that aluminium be pro- 
diiced at all. 

3. A current of 75 amperes and about 5 volts suflficed to de- 
compose the bath and to produce 105 grammes of aluminium 
in seven hours. This would be nearly 30 grammes per hour 
for each electric horse-power. 

4. A current of 80 amperes and 24 volts was passed through 
four crucibles connected in series for six hours, using a bath 
of I part cryolite and 3 parts sodium chloride with 450 
grammes of lead in each crucible. The crucibles were heated 
regularly to a moderate temperature. There were obtained 
altogether 250 grammes of quite pure aluminium. This would 
be equal to 16 grammes per electric horse-power-hour. 

" A large number of similar experiments afforded a return of 
^ to I ^ lbs. of aluminium per electric horse-power per day. 


The experimental plant now in operation consists of a 40 volt 
— TOO ampere — dynamo, the current being sent through six pots 
connected in series. When the bath is completely electrolyzed 
the contents of the crucible are tapped ofif at the bottom and a 
fresh supply of melted salt poured in quickly. The lead- 
sodium alloy run off is put back into the crucibles, thus keep- 
ing approximately constant in composition and going the 
rounds continuously. With this apparatus, 3 to 4 lbs. of alu- 
minium are produced regularly per day of 12 hours. As soon 
as patents are obtained, it is the intention of the company to 
put up a plant of 50 lbs. daily capacity, which can be easily 
increased to any extent desired as the business expands." 

Professor Rogers observes in regard to the apparatus that he 
has tried various basic linings for his clay crucibles, but a paste 
of hydrated alumina, well fired, has succeeded best. Some 
"shrunk" magnesia lining, such as is used in basic steel fur- 
naces, answered well, but could not be used because of the 
amount of iron in it. Lime could not be used, as it fluxed 
readily. The carbon rods lasted 48 hours without much cor- 
rosion if protected from the air during electrolysis. Carbon 
plates and cylinders were tried, but the solid rods gave the 
best results. About 8 to 10 per cent, of aluminium can be 
extracted from cryolite containing 12.85 P^r cent. The min- 
eral used was obtained from the Pennsylvania Salt Company, 
and was called pure, but it contained 2 per cent, of silica and 
I per cent, of iron. These impurities pass largely into the 
aluminium produced, but the company hoped to be able to 
manufacture an artificial aluminium fluoride which will not 
only be purer but less costly than this commercial cryolite. 
Professor Rogers infers that pure aluminium fluoride would not 
be an electrolyte, since the resistance of the bath increases as 
the amount of other salts present decreases. 

It is useless to base any accurate estimation of the cost of 
aluminium by this process on the data given above, since they 
were only for a small experimental plant. If, however, 75 per 
cent, of the aluminium in cryolite can be extracted at the rate 


of I lb. of metal per day per electric horse-power, and the 
metal is free from lead and sodium, (a sample sent me was 
of very fair quality,) it would seem that the process was in a 
fair way to compete on an equal footing with the other electro- 
lytic processes as operated in 1889, but it has been entirely 
distanced by more recent developments. 

Dr. Hampe on the Electrolysis of Cryolite. 

Prof. W. Hampe, of Clausthal, whose name is a guarantee of 
careful and exact observations, has written the following val- 
uable information on this subject, in presenting which we will 
also give the remarks of Dr. O. Schmidt, called forth by 
Hampe's first article. 

*" The electrolysis of a bath of cryolite mixed with sodium 
and potassium chlorides, using a layer of melted copper in the 
bottom of the crucible as cathode and a carbon rod as anode, 
gave balls of melted sodium which floated on the surface and 
burnt, but scarcely a trace of aluminium. Yet here the condi- 
tions were most favorable to the production of the bronze. The 
battery used consisted of twelve large zinc-iron elements.'' 

fDr. O. Schmidt, referring to this statement of Hampe's, 
quotes an opposite experience. He fused cryolite and sodium 
chloride together in a well-brasqued crucible in the proportions 
indicated by the reaction 

AUF6.6NaF+6NaCl=Al2Cle+i 2NaF. 

At a clear red heat the bath becomes perfectly fluid and trans- 
parent, and an anode of gas carbon and a cathode of sheet cop- 
per are introduced. On passing the current the copper did not 
melt but became covered with a film of deposited aluminium, 
which in part penetrated the electrode and in part adhered to 
the surface as a rich alloy which ultimately fused ofif and sank 
to the bottom of the crucible. With a plate i to i ^ millimetres 
thick, 10 per cent of its weight of aluminium could thus be de- 

* Ghemiker Zeitung, xii., 391 (1888). fldem, xii., 457 (i 


posited ; with one 3 millimetres thick, about 5 per cent. The 
metal could be made perfectly homogeneous by subsequent 
fusion in a graphite crucible. Dr. Schmidt further remarks 
(evidently on the supposition that the reaction he gives actually 
takes place) that on thermo-chemical grounds sodium would not 
here be reduced, because while the molecule of sodium chloride 
requires 97.3 calories for its decomposition, that of aluminium 

chloride, — ^, requires only 80.4, and the current would attack 

first the most easily decomposed. He also states that the cal- 
culated difference of potential for the dissociation of aluminium 
chloride which is ^'^ = 3.5 volts, was actually observed, and the 

tension of the current must have been increased to about 4.5 
volts to bring about the decomposition of the sodium chloride.* 

Dr. Hampe's statement occasioned several other communica- 
tions, which he considers and replies to in the following 
article : f — 

" Dr. O (whose name I withhold at his own request) 

writes to me that by electrolyzing pure cryolite, using a nega- 
ative pole of molten copper, he never obtained aluminium 
bronze ; but, on the other hand, always obtained it if he used 
the mixture of cryolite and sodium chloride mentioned by Dr. 
Schmidt, and in place of the molten copper a thick stick of the 
unfused metal. A letter from R. Gratzel, Hannover, contains a 
similar confirmation of the latter observation. By electrolyzing 
a mixture of 100 parts cryolite with 150 of sodium chloride in 

* Aside from Hampe's subsequent remarks as to no aluminium chloride being 
formed, we would further point out the fact that the decomposition of a chemically 

equivalent quantity of aluminium chloride requires not — ^ = 80.4 calories, but 

^"' °° or 53.6 calories, and the calculated difference of potential is properly -^^ 

or 2.3 volts. The fact that the observed tension was 3.5 volts shows that the current 
was not strong enough to decompose the sodium chloride, as Schmidt observes, and 
the fact that this current deposited aluminium would show that the heat of formation 
of aluminium fluoride at this temperature cannot be greater than 23X3.5X6^483 
(thousand) calories. 
tChemiker Zeitung (Cothen), xiii., 29 and 49. 


a graphite crucible holding 30 kilogrammes, aluminium bronze 
dripped down from the ring-shaped copper cathode used, while 
chlorine was freely disengaged at the carbon anode. But after 
a time, long before the complete decomposition of the cryolite, 
the formation of bronze stopped — even an attacking of that 
already formed sometimes taking place. Pellets of an alloy 
of sodium and aluminium appear on the surface and burn with 
a white light. 

" These comments excited me to further research in the mat- 
ter. At first, it was necessary to consider or prove whether by 
melting sodium chloride with cryolite a true chemical decom- 
position took place, such as Dr. Schmidt supposed. If this 
were the case, the very volatile aluminium chloride must neces- 
sarily be mostly driven off on melting the mixture, and at a 
temperature of 700° to 1000° C. there could not be any left in 
it. But an experiment in a platinum retort showed that such a 
reaction positively does not occur ; for neither was any alu- 
minium chloride volatilized, nor did the residue contain any, 
for on treatment with water, it gave up no trace of a soluble 
aluminium compound. During the melting of the mixture, acid 
vapors proceeded from the retort, and a small quantity of cryo- 
lite was volatilized into the neck of the retort. Dr. Klochman 
has shown that cryolite always contains quartz, even colorless, 
transparent pieces, which to the naked eye appear perfectly 
homogeneous, showing it when examined in thin sections under 
the microscope, and on melting the mineral opportunity is 
given for the following reactions : 

SiOa 4- 4NaF = SiF, + 2Na20, 
sNa^O + Al^Fe = 6NaF + AUO,, 

as is rendered probable by the appearance of delicate crystals 
of alumina on the inner surface of the retort just above the 
fusion. The silicon fluoride probably passes away as silico- 
fluoride of sodium. 

" If cryolite is fused with such metallic chlorides that really do 


bring about a decomposition, there is never any aluminium 
chloride formed in these cases, but the sodium of the cryolite is 

exchanged for the other metal. Dr. O , to whom I owe this 

observation, fused cryolite with calcium chloride, hoping that 
aluminium chloride would distil, but obtained instead crystals 
of the calcium salt of alumina-fluoric acid ; thus, 

NaeAljFi, + aCaClj = 6NaCl + CajAl^Fi^, 

and in like manner can be obtained the analogous strontium or 
barium compounds. 

" Just as erroneous as the supposed production of aluminium 
chloride are the other arguments advanced by Dr. Schmidt, re- 
garding the reasons why sodium could not be set free. The 
self-evident premises for the propositions are lacking, viz. : that 
the two bodies compared are conductors. On the contrary, I 
have previously shown* that aluminium chloride and bromide, 
and more certainly its fluoride, belong to the non-conductors. 
It follows, then, that there can remain no doubt that on elec- 
trolyzing pure cryolite, or a mixture of it, with sodium chloride, 
only sodium will be set free at first, either from sodium fluoride 
or the more easily decomposable sodium chloride. The pres- 
ence or absence of sodium chloride is consequently, chemically, 
without significance. 

" Since the experiments with solid cathodes gave aluminium, 
while those with molten copper did not, these results being 
independent of the presence or absence of sodium chloride, the 
next attempt made was to seek for the cause of the different re- 
sults in the differences of temperature. It was found that when 
the electrolysis takes place at a temperature about the melting 
point of copper, bubbles of sodium vapor rise and burn, and any 
aluminium set free is so finely divided that it is attacked and 
dissolved by the cryolite. To explain this action of the cryo- 
lite it is necessary to admit the formation of a lower fluoride of 
aluminium and sodium, such as I have recently proven the exist- 

* Chemiker Zeitung (Cothen), xi., p. 934 (1887). 


ence of.* The solution of the aluminium takes place according 
to the following reaction: 

Al^Fe.eNaF + Al = 3 (AlF2.2NaF) . 

If the electrolysis takes place at a temperature so low that the 
sodium separates out as a liquid (its volatilizing point is about 
900°), large globules of aluminium will be produced, on which 
the cryolite seems to exert no appreciable action. Nevertheless, 
the yield of aluminium is much below the theoretical quantity 
set free. Since pure copper melts at 1050°, and aluminium 
bronze at 800°, the copper electrodes can remain unfused in the 
bath while the bronze melts off as it forms, while the tempera- 
ture can be low enough to keep the sodium in the liquid state. 
By mixing sodium or potassium chlorides with the cryolite, the 
melting point is lowered, or at a given temperature the bath is 
more fluid, and so easier to work. When there is not enough 
aluminium fluoride present in the bath to utilize all the sodium 
liberated, the excess of sodium may form an alloy with some 
aluminium, and rising to the surface, burn to waste. Since cry- 
olite always contains silica, as previously explained, the bronze 
thus obtained is always rendered hard with silicon, and is not of 
much value commercially.'' 

Hall's Process. 
Although Hall's patents were not issued by the patent office 
until 1889, yet they were applied for in the middle of 1886, and 
a commercial-sized plant was put up to work the process in 
1888. Mr. Hall really discovered the principle on which his 
process is based, and made some aluminium by it, on a small 
scale, in February, 1886. 

*Chem. Zeit. (Cothen), xiii., p. I (1879). Hampe melted together aluminium and 
sodium fluorides in the proportions of one molecule of the first to four of the second, 
and obtained what is apparently a lower fluoride than cryolite, in which aluminium 
cannot be otherwise than diatomic, since analysis gives it the formula AlF2.2NaF. 
This salt is similar in appearance and properties to cryolite. As there are still some 
doubts, however, about this compound, the above explanation of the solution of alu- 
minium by the cryolite need not be accepted as final. 


The inventor of this process, Charles M. Hall, of Oberlin, 
Ohio, born in 1863, is a graduate of Oberlin College, and while 
yet a student spent considerable time in experimenting on iso- 
lating aluminium with the electric current. Less than a year 
after leaving college, he conceived the following idea, which I 
quote in his own words : " It occurred to me that if I could find 
some stable solvent for alumina itself, which at a reasonable 
and practicable temperature would dissolve the alumina by the 
mere mingling of it with the solvent and allow the alumina so 
dissolved to be electrolyzed out of it, leaving the solvent unaf- 
fected, this would possibly be the best process which could be 
devised for the manufacture of aluminium by electrolysis." 
Many different salts were tried, to see if they possessed the 
property required, among them the fluorides of calcium, mag- 
nesium, sodium and potassium. These latter salts were the 
only ones which gave any encouraging results, but the first two 
were too hard to fuse, and they all dissolved only small quan- 
tities of alumina. Hall next tried cryolite, the natural com- 
pound of aluminium and sodium fluorides, and found that it 
melted at a red heat and readily dissolved considerable alumina. 
This was on February 10, 1886. On applying the electric cur- 
rent to this bath he was not immediately successful ; but, at- 
tributing the failure to the presence of silica in the crucible 
lining, he tried a crucible with a carbon lining, and the experi- 
ment was entirely successful. This was February 23, 1886, and 
on July 9, of the same year. Hall applied for patents covering 
his invention. 

P. L. V. Heroult, of Paris, France, had, however, apphed for 
a United States patent for substantially the same process on 
May 22, 1886, and the patent ofifice declared an interference 
and ordered both parties to give evidence as to the date of 
invention and putting into practice. Hall was able to name 
February 23, 1886, as the date on which he had put his inven- 
tion into practical operation, while Heroult named only the 
date of his French patent for the same invention, April 23, 
1886. (See p. 386.) The result of these proceedings was 


that a patent covering the process was issued to Hall on April 
2, 1889. (Number 400,766.) This patent has claims cover- 
ing the use of the following process : 

1. Dissolving alumina in a fused bath composed of the 
fluorides of aluminium and a metal more electropositive than 
aluminium, and then passing an electric current through the 
fused mass. 

2. Dissolving alumina in a fused bath composed of the 
fluorides of aluminium and sodium, and then passing an 
electric current, by means of a carbonaceous anode, through 
the fused mass. 

In the specification of the patent, it is stated that the bath 
preferred as a solvent contains : 

Sodium fluoride 33.2 per cent. 

Aluminium fluoride 66.8 per cent. 

represented by the formula AlFj.NaF. A convenient method 
of forming this bath consists in adding to the mineral cryolite 
80.3 per cent, of its weight of aluminium fluoride, according to 
the formula : 

AlFs.sNaF + 2AIF3 = sCAlFs.NaF). 

It is further stated that the bath may be fused in a crucible 
or melting pot of iron or steel having a carbon lining, that the 
positive electrode may be of carbon, copper, platinum or any 
other suitable material, and the negative electrode a carbon 
rod or the carbon lining of the bath. When the positive 
electrode is carbon it is gradually consumed by the oxygen 
liberated, but when copper is used a coating of copper oxide 
is at once formed on it, which protects it from further oxidation 
and causes free oxygen to be liberated. The tension required 
to operate a bath is stated as 4 to 6 volts. 

The above description may be taken as summarizing Hall's 
process as developed by him at the date of application for his 
patent, July, 1886. It is the writer's opinion that credit should 
be given to both Hall and Heroult for originating this inven- 


tion, as the principle was discovered independently by each at 
very nearly the same time ; but in the subsequent development 
and practical application of the invention, Hall in America 
clearly outstripped Heroult in Europe. 

Several other patents were subsequently applied for by Hall, 
covering modifications of the original process, and these were 
all issued with the original patent in 1889.* One covers the 
use of a potassium-aluminium fluoride, AIF3.KF, instead of the 
corresponding sodium salt, since it is much more fusible. 
Also the addition of lithium fluoride to this salt. Another 
covers the use of calcium and aluminium fluorides to form the 
bath, such as 2AlF3.CaF2 or 2AlF3.3CaF2. Since this bath is of 
higher specific gravity than molten aluminium, it is recom- 
mended to add to it enough of the salt AIF3.KF to lower the 
specific gravity below that of the aluminium. This is unneces- 
sary if alloys of aluminium are to be made, in which case the 
metal to be alloyed is made the negative electrode and the 
alloy formed will be sufficiently heavy to sink. For making 
alloys, the addition of barium fluoride to the bath is recom- 
mended, as its high specific gravity is no inconvenience and it 
is more fusible than the calcium or strontium fluorides. In 
another of these patents, the bath specified has the formula 
2(AlF3.3NaF) + aAlFj-CaFj, being made by melting together 
cryolite, aluminium fluoride and fluorspar. When the bath 
becomes clogged, by the settling out of a mushy deposit which 
stops the current, the addition of 3 or 4 per cent, of anhydrous 
calcium chloride is recommended as clearing or liquefying the 
bath ; this addition is also said to lessen the disintegration of 
the carbon anodes used. 

As at present conducted, the process is worked according to 
the original patent applied for in 1886, it having been found 
that almost all the difficulties which the subsequent modifica- 
tions were designed to overcome are entirely obviated by regu- 
lar working on a large scale. It has been concluded from 

* Numbers 400,664; 400,665; 400,666; 400,667; 400,766. 


subsequent experience that almost all the difficulties met at 
first were the results of the varying composition of the bath 
and its irregular temperature, and were inherent in attempts to 
work on a small scale. 

From June, 1887, to the middle of 1888, Mr. Hall experi- 
mented on his process at the Cowles' Bros, works at Lockport, 
N. Y. His results here, working on a small scale, were not 
sufficiently economical to induce that company to continue the 
work. Mr. Hall, however, was confident that if the process 
could be started on a large scale, aluminium could be made at 
a cost very much below its selling price at that time. Leaving 
Lockport, Mr. Hall went to Pittsburgh, and there, through the 
influence of Mr. Romaine C. Cole, was successful in interesting 
several capitalists, who contributed $20,000 in cash to put up a 
works, and in September, 1888, organized the Pittsburgh Reduc- 
tion Company, with a capital stock of $1,000,000. The con- 
struction of the plant was immediately begun, and in Novem- 
ber, 1888, aluminium was being produced at the rate of 50 
pounds per day, which found a ready sale at $5 per pound, 
leaving a handsome profit. 

This plant was situated in Pittsburgh, and was operated by 
steam-power, using coal for fuel. The machinery consisted of 
two ordinary tubular boilers of sixty horse-power each, and a 
Westinghouse " automatic " engine of 125 horse-power, running 
two shunt-wound Westinghouse dynamos. These dynamos, 
running at 1,000 revolutions, gave each a current of 1,000 
amperes at 25 volts tension, and being coupled in parallel gave 
a current of 2,000 amperes at full speed, or an average of 1,700 
to 1,800 amperes at 16 volts on steady runs. This current was 
passed through two reducing pots, coupled in series. These 
pots were of cast-iron, 24 inches long, 16 inches wide and 20 
inches deep, lined inside with a layer of hard-baked carbon 3 
inches thick. From a copper bar above were suspended, by 
^ inch copper rods, 6 to 10 three-inch carbon cylinders, each 
originally 15 inches long. Each pot. held 200 to 300 pounds 
of the electrolyte. 


It was found that with apparatus and current of this size, 
outside heating was not necessary; the heat generated by the 
passage of the current being sufficient to keep the bath fused 
at the proper temperature. Larger pots have since been used, 
which will be described later, but as the operation in them is 
exactly similar to that in the smaller ones, we will describe the 
process at once. 

The carbon lining of the pot forms the negative electrode, 
connection being made to the iron casing. The fluoride salts 
to form the bath are placed in the bottom of the cavity, the 
carbon positive electrodes lowered until they touch the carbon 
lining, and the current turned on. After a few minutes the 
salt is melted and when red-hot in the vicinity of each carbon, 
more salt is thrown around the carbons, or each carbon moved 
up a little. In an hour or two sufficient salt has been melted to 
form the bath, and the carbons have been brought out of direct 
contact with the lining. The bath is now ready to dissolve 
alumina, which is first sprinkled on top of the bath, and after 
it has become hot and perfectly dry is stirred in. 

The fumes which were arising from the decomposition of the 
fluorine salts immediately cease, the voltage absorbed by the 
bath falls, the number of amperes passing increases, and the 
electrolysis of the dissolved alumina commences at once. 
Powdered carbon (the stub ends of electrodes ground up) is 
now placed on top of the bath in a layer about an inch thick, 
serving to keep in the heat and so reduce the loss by radiation. 
The carbon electrodes are clamped to a copper bar extending 
the whole length of the pot (see Fig. 38), and if one part of 
the bath becomes cooler than it should be, the carbons are 
crowded together at that point and soon heat it up. In prac- 
tice, the ends of the carbons are kept about 0.5 to i inch from 
the lining, and each one carries about 250 to 300 amperes. If 
more current than this passes through a carbon, as might happen 
if it short-circuited, the copper rod becomes red-hot at its con- 
nection with the carbon. As soon as this is observed, the 
workman raises it slightly and so stops the extra current. The 



carbons wear down gradually until only 3 or 4 inches long (the 
depth of liquid bath is not over 6 inches) and must then be 
replaced. The consumption of carbons in regular running is 
less than one pound per pound of aluminium produced, show- 
ing that they are burnt according to the reaction 

AI.O, + 3C = 2AI + 3CO. 

in which 102 parts of alumina would require 36 parts of car- 
bon to combine with its oxygen, producing 54 parts of metal. 

Fig. 38. 

Theoretically, therefore, there is required || or ^ of a pound 
of carbon to i pound of aluminium produced. The smal'l 
amount over this actually used is consumed by the air with 
which they inevitably come in contact. 

In these small pots, a tension of 8 volts is required to oper- 
ate each, at an amperage of 1800. If the voltage rises to 10 
or more, it indicates that the alumina in the bath has become 
exhausted, as is also shown by dense, irritating fluoride fumes. 


The workman at once stirs in some of the alumina which is 
always kept on top of the carbon covering, and the voltage 
immediately falls. A delicate electrical instrument, set to a 
certain voltage, gives immediate notice to the workman when 
the bath is " out of ore." In regular working no fumes arise, 
the carbon covering entirely conceals the red-hot fluid bath 
beneath, and one can only convince himself that the bath is 
really in operation by approaching and noticing the heat 
radiated. When aluminium has accumulated for several hours, 
it is ladled out with iron ladles well-rubbed with chalk or 
plumbago. This operation gives rise to some fumes, and also 
causes some of the electrolyte to be dipped out with the metal. 
It has been suggested that a tap-hole should be provided, but 
there was found great difficulty in keeping the pot tight when 
this was tried, and the expedient of syphoning out the metal 
has been found more practicable. 

The crude ingots thus poured are re-melted in large crucibles ; 
the re-melting plant is shown in Fig. 39. 

The alumina used in this plant was from the German manu- 
turer, Bergius, at Goldschmeiden, near Breslau, Silesia. As im- 
ported, it contained about 33 per cent, of moisture, which is 
driven off in a drying furnace. The cost of alumina was, in 
1890, five cents a pound delivered in Pittsburgh, which would 
make the dried alumina cost 7.5 cents per pound plus the cost 
of calcining, which is not over 0.5 cent. If absolutely pure, 
alumina should contain 52.94 per cent, of aluminium, but as 
this calcined alumina usually contains about i per cent, of im- 
purities, chiefly silica, there is only a little over 52 per cent, of 
metal in it. An accurate account of several months' running 
showed that the Hall process extracts a fraction over 50 per 
cent, of metal from it. The alumina to furnish one pound of 
aluminium therefore cost in 1890 about 16 cents. Since then, 
the alumina is calcined in Germany before shipment, saving in 
freight and customs dues, and can probably be laid down now 
in New York at S cents per pound, making a cost of 10 cents 
per pound of aluminium. There is, besides this, a small waste 



of bath material, which costs in the neighborhood of 7 cents per 
pound to make, but since this does not in actual running 
amount to over 2 pounds per 100 pounds of aluminium pro- 
duced, this item of expense is quite small. 

The output of these small pots was about one pound of 
metal an hour from each, and they were kept in continued ope- 
ration for several weeks at a time. If pure alumina is used, 

Fig. 39. 

,, "^^-^ /W''. 

pure aluminium is produced ; that made from Bergius' alumina 
averaging over 98 per cent. pure. 

When a bath has been put into operation, the metal first 
produced is contaminated with the silicon and iron present in 
the bath material, but in a few days these are completely elim- 
inated, and the bath material is almost chemically pure save for 
the impurities introduced in the alumina. 

If calcined, selected raw bauxite is used in the pots, the 
metal produced averages 94 to 96 per cent, pure, and is sold for 


use in iron and steel. Two analyses of such second quality- 
metal gave : 

Aluminium 94.16 9S'93 

Silicon 4.36 2.01 

Iron 1.48 2.06 

As the material used only costs about i cent per pound 
ready to put into the pots, and about 2^ pounds make one of 
metal, this second quality metal costs about 7 cents less per 
pound than the first quality. 

An estimate of the cost of producing aluminium in this plant 
in 1889, at the rate of fifty pounds per day, is as follows: 

100 lbs. calcined alumina @ 8 cts =. $8.00 

Fluorides for bath @ 7 cts = .50 

Carbons = i.oo 

50 horse-power engine = 15.00 

2 engineers @ $3.00 per diem = 6.00 

6 workmen @ ;j2.oo " = 12.00 

Superintendence, office expenses, etc .= 10.00 

Interest on plant, rent, etc ^ 7.50 

Cost of about 50 lbs. of aluminium $60.00 

In the early part of 1890 the plant was greatly enlarged. 
The additions consisted of three Babcock and Wilcox boilers of 
200 horse-power each, fired by natural gas ; two 200 Jiorse- 
power Westinghouse compound engines ; two shunt-wound 
Westinghouse dynamos running 325 revolutions per minute, 
connected in parallel and giving a total current of 5000 amperes 
at 50 volts tension. The conductors for this current were two 
copper bars, 6 inches by 0.5 inch, giving a total cross-section 
of 6 square inches. This gives a little over 800 amperes per 
square inch section. This current operated five large pots con- 
nected in series, averaging 9 volts to a pot. These large pots 
are of one-quarter inch wrought iron, 5 feet long, 2% feet wide 
and 2 feet deep, with a carbon lining 4 to 6 inches thick, and 
supported on bricks to keep the bottom cool. Above are two 
copper bars to which a double row of carbon electrodes are 
clamped, there being 10 on each row or 20 in all. Each of 


these pots produced 60 to 70 pounds of aluminium a day, mak- 
ing the production of the entire plant in 1890 about 400 
pounds per day. The cost of aluminium, working on this scale, 
could scarcely have been over $0.75 per pound, since the 
metal made here was sold during 1891 for $1.00 per pound. 

During 1891, in order to obtain room to extend and cheapen 
fuel, the entire plant was moved to New Kensington, Pa., on the 
Allegheny River 18 miles from Pittsburgh. At this place the 
capacity of the plant has been greatly increased, there being ap- 
proximately 1500 horse-power employed, and the output having 
been increased to lOOO pounds per day in 1893 and to nearly 
2000 pounds per day in 1894. Since the selling price has been 
reduced to $0.50 per pound, it is evident that the present 
cost cannot be much over $0.40, if so high. Captain A. E. 
Hunt, president of the company, is authority for saying that as 
the process is operated on an increasing scale, and more exper- 
ience is had in running it, the cost in the future will approxi- 
mate $0.20 per pound, the lowest limit by this process. 

In July, 1890, a plant to work Hall's process was put into 
operation at Patricoft, Lancashire, England, with a capacity of 
about 300 pounds of aluminium a day. This plant was in 
operation until 1894, when competition from Switzerland com- 
pelled it to close. 

During 1894, the United States Company have been con- 
structing a larger plant at Niagara Falls, being furnished power 
by the Niagara Falls Power Co. A contract has been made to 
take 6000 horse-power, and the plant, so far constructed, is as 
follows : * 

A 5000 horse-power turbine, in the Power Company's power- 
house, runs a 5000 horse-power, low frequency, alternating 
current Tesla dynamo. This machine, at full speed, gives two 
currents of 2500 volts by 775 amperes each, alternating 50 times 
per second. The exciting circuit is continuous, obtained from 
the main circuit by a rotary transformer, and amounts to only 
0.2 per cent, of the main current. The total current of 2500 

*This plant was put into operation on August 26, 1895. 


volts by 1550 amperes is conducted through an underground 
tunnel one-half mile to the aluminium works, through copper 
rope conductors, with a total section of 4.3 square inches, with 
a drop in potential of 12 volts. At the works, the current is 
first passed through a set of stationary transformers, each of 
250 horse-power, in which the current is reduced from 2487 
volts to 115 volts. The alternating current of 115 volts next 
passes to rotary transformers, each of 500 horse-power and 
taking the current from two stationary transformers. In these 
the alternating current of 115 volts is changed to a direct cur- 
rent of 160 volts, the quantity of current furnished by each 
rotary transformer being 2500 amperes. These currents of 
2500 amperes by 160 volts are conducted to the pot room on 
copper bars with 1.25 square inch section, each passing through 
a separate switch-board, on which are shunt ammeters and vol- 
meters, and then passed to the pots. The buildings are all of 
iron, the main reduction room being 180 by 85 feet. 

It is not difScult to estimate what the output of the two works 
will be. At Pittsburgh each horse-power of electrical energy 
produced about 1.2 pounds of aluminium per day, at which rate 
a 5000 horse-power plant should make 6000 pounds, equiva- 
lent in round numbers to 1000 tons a year, the present value of 
which is $1,000,000. 

Reactions and efficiency of the process. — Taking the first small 
plant, a current of 1800 amperes in two vats should theoreti- 
cally produce 56.4 pounds of aluminium per day. As 50 
pounds were obtained, on an average, the efficiency shown in 
this direction is nearly 90 percent. As only about 2.5 volts 
out of the 9 volts absorbed by the bath, is used in decomposing 
the alumina, wc have about 30 per cent, of the current thus 
utilized. The efficiency over all is therefore 90 per cent, of 30 
per cent.=27 per cent. This means that the work of separat- 
ing aluminium from oxygen represents 27 per cent, of the actual 
energy of the current. The difference between 9 and 2.5, or 
6.5 volts, is absorbed in the conduction resistance of the bath, 
being converted into heat. This would give a heat develop- 


ment of 6.25 X 1800x0.00024= 2.7 calories per second, or 
9720 calories per hour. To this we may add the heat of 
oxidation of the carbon to carbonic oxide, and we have the 
total heat generated in the bath. A balance would show as 
follows : 

Heat Developed per Hour. 

By electrical resistances 9720 calories. 

Oxidation of the carbon 960 " 

Total 10680 " 

Heat Distribution per Hour. 

Heating 2 pounds alumina to 900° C 250 calories. 

Withdrawn with the aluminium 17S " 

Lost by radiation and conduction 10265 " 

Total 10680 

As to the real reaction in the bath, by which aluminium is 
liberated, several metallurgists have taken the ground that the 
aluminium fluoride present is the substance primarily disso- 
ciated by the current, and that the fluorine set free at the anode 
immediately acts on the alumina present in solution and re- 
forms aluminium fluoride, liberating oxygen, which then unites 
with the carbon. According to this view the reactions are 

2AIF3 + Current = 2AI + 6F. 
MO3 + 6F = 2AIF, + 3O. 

3C + 3O = 3CO. 

This theory must assume either one of two things ; first, 
that the alumina added never becomes an integral part of the 
bath, but remains suspended mechanically; or, second, that if 
the alumina really is part of the bath, it is a stronger com- 
pound than aluminium fluoride, and therefore, the current 
decomposes the latter. This theory, however, involves several 
contradictions. If, for instance, alumina is really a stronger 
compound than aluminium fluoride, how could the gaseous 


fluorine decompose alumina, forming aluminium fluoride? 
Since gaseous fluorine can decompose alumina, as was proven 
by Moissan, it follows that alumina is therefore a weaker com- 
pound than aluminium fluoride, and that when an electric cur- 
rent is passed through a bath containing these two compounds 
the alumina is necessarily the first to be decomposed. The 
only refuge left for this theory is, then, to maintain that the 
alumina is not truly dissolved in the bath, but is only me- 
chanically suspended. This, however, is contrary to the way 
in which the bath behaves. Absorption of the fluorine could 
not possibly be as perfect in that case as it really is in practice. 
Besides, an experiment of Mr. Hall's settles the case, it seems 
to me, beyond question. On passing the current through an 
anode of clean copper, it becomes immediately coated with a 
layer of copper oxide. If fluorine were at all set free, it would 
be liberated in immediate contact with the anode, and copper 
fluoride must necessarily have been formed ; but no trace of 
that salt is observed. It thus appears impossible to me that 
the electric current does anything but dissociate oxygen from 

It has been suggested that the alumina is dissolved in the 
bath by forming aluminium oxy-fluoride, and that this is the 
substance electrolyzed. This may or may not be true, but there 
is up to the present time no positive proof for the theory, and 
several circumstances against it. No chemist has as yet iso- 
lated and analyzed any such salt. If such salt is formed, the 
larger the proportion of aluminium fluoride in the bath, the 
more alumina it ought to dissolve. But cryolite contains 40 
per cent, of aluminium fluoride, and dissolves at least 25 per 
cent, of its weight of alumina to a clear solution, while the salt 
AlFg.NaF, containing 66 per cent, of aluminium fluoride, and 
made by adding aluminum fluoride to cryolite, dissolves barely 
10 per cent. Further, no evolution of heat or other signs of 
chemical action are observed when dissolving alumina in the 
bath. If such a salt exists in solution, the elements set free at 
the anode must be oxygen and fluorine, but it is already shown 


that no fluorine can be detected there even when using a cop- 
per anode. So, while the theory of an oxy-fluoride may be 
true, and facts unknown to us now may reconcile these appa- 
rent contradictions, yet at present there is no positive proof to 
give in its favor. 

In connection with this process, the following determinations 
by the writer of specific gravities, when cold and melted at a 
red heat, of the substances used in the bath, are of interest : * 

Cold, Molten 

after fusion. at a red heat. 

Commercial aluminium 2.66 2.54 

Commercial cryolite from Greenland 2.92 2.08 

Cryolite containing' all the alumina it can 

dissolve 2.90 2.35 

Cryolite-faluminium fluoride forming the 

salt AlFj.NaF (Hall's preferred bath) .. . . 2.96 1 .97 

Same as above, but containing all the alu- 
mina it can dissolve 2.98 2.14 

SaltAlF3.KF 2.35 

Powdered calcined alumina 3.60 

H'eroiMs Processes. 

P. L. V. Heroult, of Paris, France, patented in 1886 the fol- 
lowing process, t Alumina is fluxed with cryolite, the bath be- 
ing contained in a carbon crucible serving as the negative 
electrode, which is placed inside a larger crucible and the space 
between filled with graphite. (Fig. 40.) The positive electrode 
is of carbon and dips into the fluid material. The negative elec- 
trode is of carbon, and connects with the graphite filling between 
two crucibles. A cover having holes for the electrodes is 
placed over the whole, and a layer of clay and loam packed on 
top of the cover. A current of only 3 volts will produce de- 
composition, and the alumina is alone decomposed, the cryolite 
remaining unaltered. The aluminium collects in the bottom of 
the crucible, and alumina is added from time to time to renew 
the bath. To prepare alloys, a negative electrode is made of 

* Journal of the Franklin Institute, July, 1894, p. 51. 
t French patent, 175,711; April 23, 1886. 



the metal to be alloyed, and dips into the bath, and as the 
current passes the alloy is formed and falls melted to the bottom 
of the crucible. A convenient strength of current for a crucible 
20 centimetres deep by 14 centimetres in diameter inside, with 
a carbon anode 5 centimetres in diameter, is found to be 400 
amperes, with an electro-motive force of 20 to 25 volts. 

The idea or principle involved in the above is exactly simi- 

FiG. 40. 

lar to Hall's process, and when Heroult applied for United 
States patents in 1886 the two claims interfered, and the evi- 
dence given in the patent ofHce showed Hall to be the prior in- 
ventor, so that the process in this country belongs entirely to 
Hall. (See Hall's process, p. 373.) It is evident that the pro- 
cess was discovered by these two inventors, on opposite sides 
of the ocean, at very nearly the same time, and entirely inde- 
pendent of each other. Hall, however, kept on improving his 


process until, in 1888, it was working on an industrial scale. 
Heroult, on the other hand, virtually abandoned his process 
for the time, and did not make any pure aluminium by it on a 
commercial scale until the end of 1889, at which time Hall had 
been selling aluminium for a year and Hall's patents had been 
issued over six months. It appears very much as if Heroult 
was not aware of the possibilities of his process until Hall's suc- 
cess showed the way. 

Meanwhile, Heroult had devised another method of produc- 
ing aluminium alloys, and successfully established it on a com- 
mercial scale. This process is as follows : * Pure alumina is 
melted by the intense heat of a powerful electric current, 
and then electrolyzed by the same current, using melted 
copper beneath the alumina as the cathode. The product 
is aluminium bronze. Alumina without copper or iron as 
a cathode cannot be electrolyzed to produce pure aluminium, 
as the metal produced would rise to the surface and be volatil- 
ized by the intense heat. 

The Societe Metallurgique Suisse, which acquired both of 
H^roult's patents, put up a plant to work the alloy process on 
a commercial scale in July, 1888. This firm had experimented 
some time with Dr. Kleiner's process, but abandoned it about 
the middle of 1887, to try H^roult's process, and with such 
successful results that the plant about to be described was de- 
cided on. The instalment consisted of two large dynamos con- 
structed especially for this work by the Oerliken Engineering 
Company, and directly coupled to a 300 horse-power Jonval 
turbine situated between them and mounted on a horizontal 
shaft. A separate dynamo of 300 amperes and 65 volts, driven 
by a belt from a pulley upon the main shaft, is used to excite 
the field magnets of the two large machies. These large dy- 
namos were originally intended to give each a current of 6000 
amperes at 20 volts electro-motive force when running at 180 
revolutions per minute, but sufficient margin was allowed in the 

* French patent, 1 70,003, April 15, 1887. English' patent, 7,426 (1887). U.S. 
patent, 387,876, August 14, 1888. German patent, 47,165, December 8, 1887. 


Strength of the field to be able to work to 30 volts. They have 
even worked up to 35 volts on unusual occasions without any- 
undue heating. It happens sometimes that the end of the anode 
touches the molten cathode, producing a short circuit, when the 
current will suddenly rise to from 20,000 to 25,000 amperes, 
without, however, damaging the machine. 

The main conductors are naked copper cables, about 10 cen- 
timetres diameter, and no special precautions are taken to insu- 
late them, since the current is of comparatively low potential, 
and a leakage of 100 amperes more or less in such a large cur- 
rent is too insignificant to take the trouble to avoid. An am- 
peremeter is placed in the main circuit, its dial being traversed 
by an index about i metre long, which is closely watched by 
the workman controlling the furnace. 

The furnace or crucible first used consisted of an iron box 
cast around a carbon block, the iron, on contracting by cool- 
ing, securely gripping the surface of the carbon on all sides, and 
thus insuring perfect contact and conduction of the current 
from the cathode inside. This method was found only suitable 
for small crucibles, and the next furnace was built up of carbon 
slabs held together by a strong wrought-iron casing. The in- 
terior depth of the crucible was 60 centimetres, length 50 and 
breadth 35 centimetres, which would permit the introduction of 
the carbon anode and leave a clear space of 4 centimetres all 
around it horizontally. At the bottom of the cavity is a passage 
to a tap-hole, D (Fig. 41), closed by the plug E, which is from 
time to time withdrawn to run ofT the alloy. The carbon 
anode, F, is suspended vertically above the crucible by pulleys 
and chains, which permit it to be raised and lowered easily and 
quickly. This anode is built up of large carbon slabs laid 
so as to break joints, and securely fastened together by car- 
bon pins. The whole bar is 250 centimetres long, with a 
section 43 by 25 centimetres, and weighs complete 255 kilos. 
The conductor is clamped to the anode by means of the cop- 
per plates, G. The crucible is covered on top by carbon slabs, 
H, H, 5 centimetres thick, leaving an opening just large enough 



for the anode to pass through. The openings, y, y, closed by 
the lids, K, K, serve for introducing fresh copper and alumina. 
The materials used have just been mentioned. Electrolytic 
or Lake Superior copper is used, the former being perhaps pre- 
ferred if from a good manufacturer. The alumina is bought as 
commercial hydrated alumina, costing in Europe 22 francs per 

100 kilos (2 cents per lb.). This is, of course, calcined before 
using, each 100 kilos furnishing about 65 kilos of alumina. 
Corundum can be substituted for the artificial alumina ; some 
from North Carolina was tried, and is said to have given even 
more satisfactory results. Commercial bauxite has been used, 
but since it contains more or less iron its use is confined to the 
manufacture of ferro-aluminium. It is very cheap in Europe, 
and requires no other preparation than simple calcining. 

The operation is begun"! by placing copper, broken into 
rather small pieces, in the crucible. The carbon anode is then 
approached to the copper, which is quickly melted by the cur- 
rent. The bath of fluid copper then becomes the negative 


pole, and ore is immediately fed into the crucible. It also is 
soon melted and floats on top of the copper. The electrolysis 
now proceeds, care being taken that while the anode dips into 
the molten ore, it does not touch the molten cathode. Par- 
ticular stress is laid on the economy of keeping the distance 
between the electrodes small, the reason given being that "the 
space between, being filled with a layer of badly-conducting 
molten ore, offers a resistance which increases with the dis- 
tance ; and although resistance is necessary, in order that the 
current should produce heat, it is not economical to have more 
heat than is necessary to melt the ore — the work of separating 
the metal from the oxygen being chiefly done by the electro- 
lytic action of the current, and not by the high temperature." 
In practice, this intervening space is not over 3 millimetres 
(one-tenth of an inch). The workman in charge, by watching 
the indications of the amperemeter, is enabled to maintain the 
anode at its proper distance without any difficulty ; it is pro- 
posed to do this regulating automatically by means of an 
easily-constructed electrical device. The oxygen liberated 
gradually burns away the anode, it being found that about one 
kilo of the anode is consumed for every kilo of aluminium pro- 

After the operation commences, the alumina and metal are 
introduced alternately in small quantities at frequent and 
regular intervals, and the alloy is tapped out about every 
twelve hours. The only wear to which the crucible is sub- 
jected is from the accidental admission of small quantities of 
air; this waste is scarcely appreciable. All oxygen evolved 
from the bath is evolved in contact with the anode and burns 
it. When the anode has worn down until too short for further 
use, it is replaced by another, the pieces of carbon left being 
utilized for repairing, covering or building up the crucible. 
The operation is kept up night and day, and it is generally 
more than a day after starting before the crucible is thoroughly 
up to its maximum heat and work. Two reliefs of five men 
each operate the plant, one to superintend, one to prepare 


and dry the alumina, a third to control the working of the cru- 
cible by regulating the anode, a fourth to feed ore and metal 
into the crucible, and the fifth to take care of the machinery, 
prepare anodes, crucibles, etc. All five work together to 
replace an anode or tap the crucible. A part of each tapping 
is analyzed, to determine its percentage of aluminium. It is 
the aim to produce as rich a bronze as possible at the first 
operation (over 42 per cent, of aluminium has been reached), 
and its subsequent dilution to any percentage desired is done 
in any ordinary smelting furnace. 

The average current supplied the crucible is 8000 amperes 
and 28 volts, requiring an expenditure of a little over 300 horse- 
power in the turbine. Starting cold it required, in one instance, 
36 hours to produce 670 kilos of aluminium bronze containing 
18.3 per cent, or 122.67 kilos of aluminium. Taking the cur- 
rent as 300 electric horse-power, this would be a return of 1 1 
grammes per hour or 0.264 kilos (0.6 lbs.) per day for each 
electric horse-power. It is claimed by Mr. Heroult that the 
furnace takes several days to attain its full efficiency, and that 
when it does so the above charge can be worked in 12 hours, 
which would triple the above production per horse-power. This 
claim is backed by figures as to 271 hours of actual operation, 
during which time the crucible cooled several times, but the 
average over the whole period was 22^ grammes per hour or 
0.544 kilo (1.2 lbs.) per day for each horse-power (163 kilos 
of aluminium per day, total production). During actual opera- 
tion at full efficiency, Mr. Heroult claims to get 35 to 40 
grammes of aluminium per horse-power-hour, which would 
mean 11 to 15 horse-power-hours per pound of aluminium, or 
1.75 to 2.1 lbs. of aluminium per horse-power per day. 

An idea of the percentage of the useful effect derived from the 
current may be had very easily by considering that i electric 
horse-power = 750 Watts = 644.4 calories of heat per hour. 
(See p. 302.) As each gramme of aluminium evolves 7.25 
calories in forming alumina, the production of i electric horse- 
power in I hour (if its energy were utilized solely for separat- 


ing aluminium from oxygen) would be 88.88 grammes. 
Therefore the heat energy of the current is amply sufificient to 
account for all the alumina decomposed, leaving over the heat 
produced in the crucible by the union of oxygen with the car- 
bon anode. Looking at the other side of the question, the 
electrolytic action, we can easily calculate from the strength 
of the current what it could perform. A current of 8000 
amperes can liberate 2.68 kilos of aluminium per hour,* ac- 
cording to the fundamental law of electro-deposition. If, then, 
from the figures given, there were actually produced 3.3 kilos 
and 6.8 kilos per hour, and 10.5 to 12 kilos are claimed when 
up to the full efficiency, it is impossible that more than a frac- 
tion of the aluminium is produced by electrolytic decomposition 
of alumina, and the claim that the process is essentially electro- 
lytic is without foundation. Similar calculations with the data 
given with regard to Cowles' process will lead to exactly 
similar conclusions, viz : that the absolute energy of the 
current, if converted into its heat equivalent, is many times 
more than sufficient to account for the decomposition of the 
alumina on thermal grounds, but the amount of current used 
will not suffice to explain the decomposition of the alumina as 
being electrolytic. Therefore, in both these processes, the 
oxygen is abstracted from alumina by carbon, the condition 
allowing this to take place being primarily the extremely high 
temperature and secondarily the fluidity of the alumina. The 
presence of copper is immaterial to the reduction, as is clearly 
shown in the Cowles process. 

In November, 1888, the AUgemeine Electricitat Gesellschaft 
of Berlin united with the Societe Electrometallurgique Suisse 
in purchasing the Heroult Continental patents and organizing 
the Aluminium Industrie Actien Gesellschaft, with a capital of 
10,000,000 marks. Dr. Kiliani, the well-known electro- 
metallurgical expert, was made manager of the works. On his 
taking charge, a much larger plant was begun, including 
foundries and machine shops for casting and utilizing ten tons 
* N. B. Only one furnace was used on the circuit. 



of bronze which it was intended to produce daily. This plant, 
consisting of 8 crucibles, was put into operation in 1891, with 
a capacity of 1000 kilos of alloyed aluminium daily. 

About this time, Heroult in co-operation with Kiliani took 
up again his process for producing pure aluminium, stimulated 
thereto, no doubt, by the success of the Hall process in 
America. They found that by modifying slightly the shape of 
their crucibles .and using a bath of cryolite to dissolve the 
alumina, the operation could be practically carried on in 
their alloy type of furnace. The apparatus used for producing 
pure aluminium is shown in Fig. 42. It will be noticed that 

Fig. 42. 

the carbon lining is not needed so thick as in the alloy fur- 
naces, since the temperature' is so much lower; also that the 
entire lining is not taken as the negative electrode, but a nega- 
tive pole of carbon is inserted] through the bottom of the 
crucible. This is the form of apparatus now in use at the 
Swiss works, producing pure aluminium so cheaply that the 
alloy process is nearly abandoned. 

The plant at the Rhine Falls, as now operated, consists of- 


the 300 horse-power turbine used since 1888, two 600 horse- 
power turbines also of the Jonval type, put into operation in 
1 89 1, and five other 610 horse-power turbines, four of which 
were put into operation in May, 1893, and the other a few 
months later. The water-rights obtained by the company in 
1889 from the Canton Schaffhausen allowed it to take from the 
Rhine 20 cubic metres (700 cubic feet) of water per second, 
which at a fall of 20 metres gives an effect of 4000 horse- 
power. The combined power of the above turbines exceeds 
this, but one is always kept at rest and only seven regularly 
run. The turbines are worked with an upward current of 
water, and the dynamos are revolved horizontally like huge 
tops above the turbine-shafts. The 600 horse-power dynamos 
are of the Oerlikon make, and at 150 revolutions per minute 
give a steady current of 7500 amperes at a tension of 55 volts. 
The current is carried by massive copper conductors to the 

This plant, working to its full capacity since the middle of 
1893, turns out daily 2500 kilos (5500 pounds) of aluminium, 
equal to 900 metric tons, a year. It is said that this amount is 
not sufficient to supply the demand for the metal, and that the 
company contemplates acquiring a large water power at Rhein- 
felden, near Basle, and erecting there an even larger plant than 
their present one. 

In 1888 the Societe Electro-metallurgique Franfaise estab- 
lished a works at Froges (Isere), 12 miles from Grenoble, to 
manufacture aluminium alloys by the Heroult process. They 
commenced selling alloys in the early part of 1889, and soon 
after commenced making pure aluminium in the same manner 
as has been described at Neuhausen. Their plant consists of 
two turbines of 300 horse-power each, with two dynamos of 
7000 amperes and 20 volts each, with an output estimated at 
3000 kilos of alloys per day, which means 200 to 300 kilos of 
aluminium. There are ten crucibles, five of which are shown 
in Fig. 43. Each crucible requires about 10 volts to run it, 
and produces i to 1.2 kilos of aluminium per hour, with an ex- 



penditure of about 2 horse-power for each kilogramme pro- 
duced per day. The output during 1893 was 50,000 kilos of 
aluminium, over 97.5 per cent. pure. 

In 1893 this company acquired a large water power at La 
Praz, near Modane, and expected to be manufacturing alumin- 
ium there in 1894. 

The United States Aluminium Metal Co. was organized to 

Fig. 43. 


work the Heroult alloy process in this country. An experi- 
mental plant, under Mr. Heroult's personal direction, was 
started at Bridgeport, Conn., in July, 1889, but the dynamo 
proved inadequate, and was burnt out. A new one was ordered 
from the Oerlikon Works, Zurich, which arrived the following 
November, and was put to work early in 1890 at Boonton, 
N. J. This dynamo was driven by water-power, and at 220 
revolutions gave a current of 3500 amperes at 35 volts tension. 
This current was sent through one crucible, shown in Fig. 44, 
producing aluminium bronzes and ferro-aluminium. This 
plant was only run for a few months, and no attempt has been 
made to enlarge it to an industrial scale. 



As the Neuhausen works are now (1895) selling aluminium 
at $0.35 per pound in large quantities, it is interesting to at- 
tempt to estimate what the cost is at present. They sell three 
quahties of aluminium, designating them as follows: — 

Flii. 44. 


Silicon. Iron. Aluminium. 

No. o 0.06 0.04 99.90 

fO.lS 0.21 99.61 
No. I '^ 

1 0.51 0.34 99.33 

N0.2 1't ''^^ ""^V 

1 3-82 3.34 92.84 
The above price is for No. i metal. In fact, the cost of these 


different grades differs only in the cost of the raw material, all 
other expenses being the same. Making calculations on No i 
grade, the cost at Neuhausen is probably as follows per kilo of 
aluminium : — 

2 kilos calcined alumina, @ fr. 0.40 0.80 francs. 

0.3 kilos cryolite, @ fr. 0.75 0.25 " 

1.2 kilos carbon electrode, @ fr. 0.30 0.35 " 

Power — 40 H. P., hours 0.15 " 

Labor and superintendence o-3S " 

General cost, repairs, etc 0.50 " 

2.40 " 

The cost of No. i aluminium at Neuhausen is therefore at 
present about 2.50 francs per kilo, or 23 cents per pound. 
The additional cost of alumina for making No. O metal is prob- 
ably 0.60 francs, and the lower cost of No. 2 metal, due to 
using raw bauxite, about 0.40 francs. One of the ofiScers of 
the company is quoted as saying that they could make No. i 
aluminium for 2 francs per kilo if working a plant producing 
10,000 kilos per day. A plant of such a size is one of the pos- 
sibilities of the next five years. 

Faure's Proposition. 
Camille A. Faure, whose process of making aluminium 
chloride is described on p. 166, proposes to obtain the metal 
therefrom by electrolysis, using carbon electrodes. M. Faure 
states that if the process is carried out on a large scale the 
chlorine set free can be utilized to form bleaching powder, and 
will thus nearly repay the whole cost of manufacturing the alu- 
minium. Patents have been applied for, covering the details of 
the electrolytic apparatus. The inventor states that he has de- 
termined on a large scale that anhydrous, molten aluminium 
chloride can be practically decomposed at 300° C. by an electro- 
motive force of 5 volts, which comprises the force required for 
actual decomposition and also that required to overcome the 
resistance of the bath. At this rate, it would require an expen- 
diture of 840 horse-power to produce 1000 kilos of aluminium 


per day, which is about twice as efficient as present processes ; 
but Mr. Faure bases his ideas of economy entirely on the value 
of the bleaching powder obtained. He is sanguine of being 
eventually able, if working on a sufficiently large scale, to make 
aluminium at five cents per pound, and is still experimenting 
in his laboratory at Grenoble. 

Winkler's Patent. 

August Winkler,* of Gorlitz, proposes to electrolyze a fusible 
phosphate or borate of aluminium. This bath is made by 
melting alumina or kaolin with phosphoric or boracic acid, the 
proportions being such that the acid is saturated ; the separa- 
tion of aluminium will not be hindered if alumina is added 
continually to combine with the acid set free. Carbon elec- 
trodes are used. 

Willson's Process. 

Mr. Willson, organizer of the Willson Aluminium Company, 
of Brooklyn, N. Y., has patented the reduction of aluminium 
alloys in an electrical furnace, as follows : The negative elec- 
trode is a graphite crucible resting on a carbon slab to which 
connection is made. The positive electrode is a hollow carbon 
rod, passing through a hole in the cover. The operation is 
similar in every respect to Heroult's alloy process (p. 390), ex- 
cepting that reducing gases, as illuminating gas, are passed into 
the crucible through the hollow anode during the operation, and 
it is stated that the anode does not touch the molten material 
beneath. Copper and corundum are charged, and alloys with 
3 to 18 per cent, of aluminium are made. It is claimed that 
the reducing gases perform the larger part of the reduction, 
and in addition prevent the wasting away of the carbon anode. 
A larger dynamo was constructed to work this process, and the 
plant erected at Spray, N. C, near the corundum deposits, but 
very little of the alloy has found its way into the market. 

* German Patent, 45,824, May 15, 1888. 


Grabau's Process* 

This process proposes the electrolysis of a bath composed of 
aluminium fluoride and caustic soda or potash or their carbon- 
ates. The particular advantages claimed are that the alumin- 
ium fluoride is procurable very pure by the inventor's chemical 
processes (see p. 172), also that the alkaline salts are easily 
procurable free from iron and silica, and if the operation is con- 
ducted in a properly-constructed crucible, very pure aluminium 
can be made. The electricity is supposed to break up the alu- 
minium fluoride, the fluorine attacking the alkali forming an 
alkaline fluoride, while the carbonic acid is driven off as gas. 
If an excess of aluminium fluoride is used, cryolite will be left 
in the bath, which can be sold as a by-product. 

It does not seem probable that this process can ever compete 
in cheapness with those based on the direct electrolysis of 
alumina, nor make purer metal than can be made by the others 
with a little extra care. 

Bucherer's Process. \ 

This inventor claims the electrolysis of the double sulphides 
of aluminium and an alkaline or alkaline-earth metal, mixed 
with molten halogen salts. The double sulphide is made by 
melting alumina with an alkaline polysulphide, sulphur and 
carbon according to the reaction 

3Na,S + AlA + 3C + 3S = Al,S3.3Na,S4-3CO. 

This double sulphide dissolves easily in a fused bath of alka- 
line or alkaline-earth chlorides or fluorides ; for instance, molten 
sodium and potassium chlorides work well either alone or to- 
gether. The bath needs only a feeble current to electrolyze it, 
and produces very pure aluminium. 

While these claims may be all true, yet it appears that con- 
siderable expense would be incurred in making the sodium 

* German Patent, 62851, June 13, 1891. 
t German Patent, 63995, 1892. 


polysulphide and in taking care of the sodium sulphide accum- 
ulating in the bath. Pure alumina would have to be provided, 
to start with, and the additional expense for chemicals is all in- 
curred to save say 50 per cent, of the electrical power used in 
other processes. It does not appear that the process will be 
as cheap as those based on the direct electrolysis of alumina. 

Another Sulphide Process. 

The Aluminium Industrie Actien-Gesellschaft of Neuhausen 
have taken out the German patent. No. 68909, in 1892. This 
patent claims the electrolysis of aluminium sulphide dissolved 
in a bath of alkaline or alkaHne-earth chlorides or fluorides, 
being kept molten during electrolysis by the electrical heat. 
The current needed for the bath is 2.5 to 3 volts if melted by 
exterior heat, 5 volts if melted by the internal heating of the 
current. As particular advantages are claimed^ 

1 . The carbon electrode is not destroyed, it being at a lower 
temperature than that at which carbon and sulphur combine. 

2. The small voltage required (theoretically, less than i volt 
is needed for decomposition). 

3. The aluminium sinks readily in the bath, avoiding short 

4. The sulphur set free can be condensed and used over in 
making the aluminium sulphide. 




No very exact classification of these numerous propositions 
can be made, since often many reducing agents are claimed in 
one general process. Where such general statements are made, 
the method will be found under the most prominent reducing 
•agent named, with cross references under the other headings. 

Reduction by Carbon without the Presence of other 


About the first attempt of this nature we can find record of 
is the following article by M. Chapelle:* — 

" When I heard of the experiments of Deville I desired to 
repeat them, but having neither aluminium chloride nor sodium 
to use I operated as follows : I put natural clay, pulverized and 
mixed with ground sodium chloride and charcoal, into an ordi- 
nary earthen crucible, and heated it in a reverberatory furnace, 
with coke for fuel. I was not able to get a white heat. After 
cooling, the crucible was broken, and gave a dry pulverulent 
scoria, in which were disseminated a considerable quantity of 
small globules about one-half a millimetre in diameter, and as 
white as silver. They were malleable, insoluble in nitric or 
cold hydrochloric acid, but at 6o° C. dissolved rapidly in the 
latter with evolution of hydrogen ; the solution was colorless, 
and gave with ammonia a gelatinous precipitate of hydrated 
alumina. My numerous occupations did not permit me to 
assure myself of the purity of the metal. Moreover, the ex- 
periment was made under conditions which leave much to be 

* Compt. -Rendus, 1854, vol. xxxviii., p. 358. 


desired, but my intention is to continue my experiments, and 
especially to operate at a higher temperature. In addressing 
this note to the Academy I but desire to call the attention of 
chemists to a process which is very simple and susceptible of 
being improved. I hope before many days to be able to ex- 
hibit larger globules than those which my first experiment 

M. Chapelle never did address any further communications 
to the Academy on this subject, and we must presume that 
further experiments did not confirm these first ones. The 
author was once called upon to examine a slag full of small, 
white metallic globules, the result of fusing slate-dust in a 
similar manner to M. Chapelle's treatment of clay. They 
proved to be globules of siliceous iron reduced from the iron 
oxide present in the slate. It is not impossible that Chapelle's 
metallic globules were something similar in composition to 

G. W. Reinar* states that the pyrophorous mass, which 
results from igniting potash or soda alum with carbon, contains 
a carboniferous alloy of aluminium with potassium or sodium, 
from which the alkaline metal can be removed by weak nitric 

The manager of an aluminium company in Kentucky claims 
to produce pure aluminium by a process which the newspapers 
state consists in smelting down clay and cryolite in a water- 
jacketed cupola reducing furnace, it being also stated that the 
aluminium is reduced so freely, and gathers under the slag so 
well, that it is tapped from the furnace by means of an ordinary 
syphon-tap. These are all the particulars which have been 
made public. As to whether this company really does make 
aluminium by any such process I am unable to assert; pure 
aluminium has been sold by this company, but that it was made 
by this company, or by any such process, is very doubtful. 

O. M. Thowless,t of Newark, N. J., proposes to prepare a 

* Wagner's Jahresb., 1859, p. 4. 

t U. S. Patent, 370,220, Sept. 20, 1887. English Patent, 14,407 (1886). 


solution of aluminium chloride by dissolving precipitated alu- 
minium hydrate in hydrochloric acid. The solution is con- 
centrated and mixed with chalk, coal, soda, and cryolite, 
and the mass resulting heated in closed vessels to a strong, 
red heat. It is also stated that aluminium fluoride may be 
used instead of the chloride. The resulting fused mass is 
powdered and washed, when it is said, that aluminium is ob- 
tained in the residue. 

According to a patent granted to Messrs. Pearson, Liddon, 
and Pratt, of Birmingham,* an intimate mixture is made by 
grinding together — 

lOO parts cryolite. 

50 " bauxite, kaolin or aluminium hydrate. 
50 " calcium chloride, oxide or carbonate. 
50 " coke or anthracite. 

These are heated to incipient fusion in a carbon-lined furnace 
or crucible for two hours, when the aluminium is said to be 
produced and to exist finely disseminated through the mass. ' 
A mixture of 25 parts each of potassium and sodium chlorides 
is then to be added and the heat raised to bright redness, when 
the aluminium collects in the bottom of the crucible. A better 
utilization of the fine powder is effected by washing it, drying, 
and then pouring fused zinc upon it, which alloys with the alu- 
minium and can be afterwards removed by distillation. If 
melted copper is used, a bronze is obtained. 

H. Bessemer, jr., of Westend,t tries to utilize the principle 
of combustion under pressure in order to get a temperature 
high enough to reduce alumina by carbon or a hydrocarbon 
gas, the resulting aluminium vapor being rapidly cooled by 
passing the products of the reaction through a small aperture 
into a large chamber, the expansion cooling and condensing 
the vapor. To conduct the process, an intimate mixture of 
alumina and carbon is made into briquettes and put into a 

* English Patent, 5,316, April 10, 1888. 
t English Patent, 12,033, J"ly 29, 1889. 


Strong iron vessel having a refractory lining and water-cooled 
externally. Air heated in regenerative stoves and gaseous 
fuel are forced into it at a pressure of three to four atmos- 
pheres. A very high temperature is produced, sufficient it is 
claimed to reduce the alumina. The products of combustion 
and metallic vapor pass out through a fire-clay tube of small 
size into a larger chamber, in which the aluminium vapor 

This process appears imaginary. Heating up the inside of 
such a large vessel, water-cooled externally, to a temperature 
sufficient to cause aluminium vapor to distil out of it, is ex- 
tremely problematical, especially with the means of heating 

Reduction by Carbon and Carbon Dioxide. 

J. Morris,* of Uddington, claims to obtain aluminium by 
treating an intimate mixture of alumina and charcoal with car- 
bon dioxide. For this purpose, a solution of aluminium 
chloride is mixed with powdered wood-charcoal or lampblack, 
then evaporated till it forms a viscous mass which is shaped 
into balls. During the evaporation hydrochloric acid is given 
off. The residue consists of alumina intimately mixed with 
carbon. The balls are dried, then treated with steam in appro- 
priate vessels for the purpose of driving off all the chlorine, 
care being taken to keep the temperature so high that the 
steam is not condensed. The temperature is then raised so 
that the tubes are at a low red heat, and dry carbon dioxide, 
CO2, is then passed through. This gas is reduced by the car- 
bon to carbonic oxide, CO, which now, as affirmed by Mr. 
Morris, reduces the alumina. Although the quantity of car- 
bonic oxide escaping is in general a good indication of the pro- 
gress of the reduction, it is, nevertheless, not advisable to 
continue heating the tubes or vessels until the evolution of this 
gas has ceased, as in consequence of slight differences in the 
consistency of the balls some of them give up all their carbon 

* Dingier, 1883, vol. 259, p. 86. German Patent, No. 22,150, August 30, 1882. 


sooner than others. The treatment with carbon dioxide lasts 
about thirty hours when the substances are mixed in the pro- 
portion of s parts carbon to 4 parts alumina. Morris states 
further that the metal appears as a porous spongy mass, and is 
freed from the residual alumina and particles of charcoal either 
by smelting it, technically " burning it out," with cryolite as a 
flux, or by mechanical treatment. 

To produce any aluminium at all by such a process and at 
such a low temperature, appears to me impossible. 

Mr. P. A. Emanuel, of Aiken, S. C, had the idea of produc- 
ing aluminium sulphide by passing carbon bi-sulphide vapor 
over aluminium sulphate, and then reducing it by carbonic 
oxide. The reactions proposed would be 

A1,(S04)«-I- 4CS, - A1,S3 + 4COS + 4SO, 
MS, + 3CO = 2AI + 3COS. 

It seems to me that the first reaction would be very likely to 
occur, because alumina alone would form the sulphide under 
the same conditions. The reduction equation, however, is very 
problematical. We do not know that carbon oxysulphide 
would be produced by these substances, but even if it might be 
formed, the reaction is thermally negative to an extent which 
would render a high temperature necessary. In fact, the heat 
deficit of the reaction is as follows : 

Calories. Calories. 

Decomposition of AljSj 124400 

Decomposition of 3CO 87000 

21 1400 

Formation of 3COS i 148100 

Deficit 63300 

The reaction may possibly occur at a high temperature, but 
I have no information that it has been experimentally accom- 

Reduction by Hydrogen. 

F. W. Gerhard * decomposes aluminium fluoride or cryolite 
* Watts' Dictionary, article " Aluminium." 


by subjecting them to hydrogen at a red heat. The aluminium 
compound is placed in a number of shallow dishes of glazed 
earthenware, each of which is surrounded by a number of other 
dishes containing iron filings. These dishes are placed in an 
oven previously heated to redness, hydrogen gas is then admit- 
ted and the heat increased. Aluminium then separates, hydro- 
fluoric acid, HF, being formed, but immediately taken up by 
the iron filings, and thereby prevented from reacting on the 
aluminium. To prevent the pressure of the gas from becoming 
too great, an exit tube is provided, which may be opened or 
closed at pleasure. This process, patented in England in 1856, 
No. 2920, is ingenious, and some one said that it yielded good 
results. The inventor, however, returned to the use of the 
more costly reducing agent, sodium, which would seem to imply 
that the hydrogen method had not quite fulfilled his expec- 

(See also Comenge's processes.) 

Quite recently the possibility of reducing alumina by hydro- 
gen has been very neatly proven by Mr. H. Warren.* He 
took a tube of fine-quality compressed lime and put into it 
alumina, made by evaporating a solution of aluminium chloride 
to dryness and calcining. He then passed pure, dry hydro- 
gen gas through the tube, heating the outside with an oxyhy- 
drogen blow-pipe flame, concentrating all the heat on one spot. 
With a slow current of hydrogen there was no reduction, but 
with a brisk current some water was formed, and there were 
found, on cooling the tube and breaking it, distinct metallic 
beads around the space of greatest heating. These were de- 
tached, and found on examination to be pure aluminium. A 
mixture of alumina and copper oxide readily gave under these 
conditions aluminium bronze. Oxides of tungsten and manga- 
nese were likewise reduced, but magnesia, silica and titanic 
oxide were not affected. 

* Chemical News, August 31, 1894, p. 102. 

408 aluminium. 

Reduction by Carburetted Hydrogen. 

Mr. A. L. Fleury,* of Boston, mixes pure alumina with gas- 
tar, resin, petroleum, or some such substance, making it into a 
stiff paste which may be divided into pellets and dried in an 
oven. They are then placed in a strong retort or tube which is 
lined with a coating of plumbago. In this they are exposed to 
a cherry-red heat. The retort must be sufficiently stong to 
stand a pressure of from 25 to 30 lbs. per square inch, and be 
so arranged that by means of a safety valve the necessary 
amount of some hydrocarbon may be introduced into the re- 
tort among the heated mixture, and a pressure of 20 to 30 lbs. 
must be maintained. The gas is forced in by a force pump. 
By this process the alumina is reduced, the metal remaining as 
a spongy mass mixed with carbon. This mixture is remelted 
with metallic zinc, and when the latter has collected the alu- 
minium it is driven off by heat. The hydrocarbon gas under 
pressure is the reducing agent. The time required for reduc- 
ing 100 lbs. of alumina, earth, cryolite, or other compound of 
aluminium, should not be more than four hours. When the 
gas can be applied in a previously heated condition as well as 
being strongly compressed, the reduction takes place in a still 
shorter period. 

Nothing is now heard of this process, and it has been pre- 
sumably a failure. It is said that several thousand dollars were 
expended by Mr. Fleury and his associates without making a 
practical success of it. We should be glad to hear in the 
future that their sacrifices have not been in vain, for the pro- 
cess has possibilities in it which may some time be realized. 

Petitjean f states that aluminium sulphide, or the double sul- 
phide of aluminium and sodium (see p. 177), may be reduced 
by putting them into a crucible or retort, through the bottom 
of which is passed a stream of carburretted hydrogen. Some 
solid or liquid hydrocarbon may be placed in the bottom of the 

* Chemical News, June, 1869, p. 332. 
tPolytechnisches Central Blatt., 1858, p. 888. 


crucible. The aluminium is said to be thus separated from its 
combination with sulphur. The powder must be mixed with 
metallic filings, as iron, and melted in order to collect the alu- 
minium. Or, metallic vapor may be passed into the retort in 
place of carburetted hydrogen. 

Messrs. Reillon, Montague, and Bourgerel* patent the pro- 
duction of aluminium sulphide (see p. 177) and its reduction 
by carburetted hydrogen exactly as above. 

Lebedeff-f fluxes alumina in a graphite crucible with fluor- 
spar, adding a double fluoride of aluminium and sodium to re- 
duce the specific gravity of the bath, exactly as in the present 
electrolytic processes. A reducing gas, as hydrogen or a 
hydrocarbon, is then to be led into the crucible. 

This method of reducing alumina has appeared so promising 
to me, that, at my suggestion. Dr. Lisle, of Springfield, Ohio, 
made some experiments of this kind, using natural gas as the 
reducing agent. A graphite crucible was lined with carbon, 
and cryolite melted in it. Alumina was then dissolved in the 
bath until the cryolite was saturated. A current of heated 
natural gas (90 parts methane to 10 parts hydrogen) was then 
passed through for several hours, the temperature being a 
bright red. No evidences of reduction were observed. The 
experiment was also modified by passing in carbonic oxide, 
with a like result. Although these experiments were fruitless, 
yet I do not think that all the po.«sibilities have been exhausted, 
and I heartily recommend this as a promising field for further 

R. E. Green, of Southall, England, | modified the experiment 
by filling a retort with charcoal and then placing the alumina 
fluxed with cryolite on top. On heating, the molten salt 
trickles down over the carbon, while a reducing gas, hydrogen 
or a hydrocarbon, is sent upwards through the retort from an 
aperture below. Mr. Green, however, spoils his process and 

* English Patent, 4576, March 28, 1887. 

t German Patent, 57768 (1889). 

J English Patent, 5914, April 6, 1889. 


incidentally shows his ignorance of the properties of aluminium, 
by recommending the addition of silica to the alumina, or the 
use of some silica to make a slag. I might have been inclined 
to believe that some aluminium could have been produced in 
this retort, provided it had a carbon lining, but the addition of 
silica would render such" a result very unlikely, or in any event, 
a worthless metal would be obtained. 

Reduction by Cyanogen. 

According to Knowles' patent,* aluminium chloride is re- 
duced by means of potassium or sodium cyanide, the former, 
either fused or in the form of vapor, being brought in contact 
with either the melted cyanide or its vapor. The patent further 
states that pure alumina may be added to increase the product 
The proportions necessary are in general — 

3 equivalents of aluminium chloride. 

3 to 9 " potassium or sodium cyanide. 

4 to 9 " alumina. 

Corbelli, of Florence,! patented the following method in 
England : Common clay is freed from all foreign particles by 
washing, then well dried. One hundred grammes of it are 
mixed with six times its weight of concentrated sulphuric or 
hydrochloric acid ; then the mixture is put in a crucible and 
heated to 400° or 500° C. The mass resulting is mixed with 
200 grammes of dry yellow prussiate of potash and 1 50 grammes 
of common salt, and this mixture heated in a crucible to white- 
ness. After cooling, the reduced aluminium is found in the 
bottom of the crucible as a button. 

According to Deville's experiments, this process will not give 
any results. Watts remarks that any metal thus obtained must 
be very impure, consisting chiefly of iron. 

Lowthian Bell J attempted to obtain aluminium in his labora- 

* Sir Francis C. Knowles, English Patent, 1857, No. 1742. 

t English Patent, 1858, No. 142. 

J Chemical Reactions in Iron Smelting, p. 230. 


tory by exposing to a high heat in a graphite crucible mixtures 
of alumina and potassium cyanide, with and without carbon. 
In no case was there a trace of the metal discovered. 

Reduction by Double Reaction. 

M. Comenge,* of Paris, produces aluminium sulphide (see p. 
177) and reduces it by heating it with alumina or aluminium 
sulphate in such proportions that sulphurous acid gas and alu- 
minium may be the sole products. The mixture is heated to 
redness on the bed of a reverberatory furnace, in an unoxidiz- 
ing atmosphere, the reaction being furthered by agitation. It 
is stated that the resulting mass may be treated in the way 
commonly used in puddling spongy iron, and afterwards 
pressed or rolled together. The reactions involved would be, 
if they occurred, 

AI2S3 + 2 AI2O3 = 6 Al + 3SO,. 
A\& + Al2(S04)3 = 4Al + 6S0,. 

It is also claimed that metallic alloys may be prepared by the 
action of metallic sulphides on aluminium sulphate ; as, for 
instance — 

Al,(S04)3 + 3FeS = Al2Fe3-h 6SO2. 

The sulphide is also reduced by hydrogen, iron, copper or 
zinc, the reactions being 

A1,S3 + 6H=2A1-|-3H2S. 
Al2S3 + 3Fe = 2Al + 3FeS. 

In the case of reduction by a metal, alloys are formed. 

Mr. Niewerth's f process may be operated in his newly in- 
vented furnace, but it may also be carried on in a crucible or 
other form of furnace. The furnace alluded to consists of three 
shaft furnaces, the outer ones well closed on top by iron covers, 
and connected beneath by tubes with the bottom of the middle 

* English Patent, 1858, No. 461, under name of J. H. Johnson. 
fSci. Am. Suppl., Nov. 17, 1885. 


one : the tubes being provided with closing valves. These 
side-shafts are simply water-gas furnaces, delivering hot water- 
gas to the central shaft, and by working the two alternately 
supplying it with a continuous blast. The two producers are 
first blown very hot by running a blast of air through them 
with their tops open, then the cover of one is closed, the blast 
shut off, steam turned on just under the cover, and water-gas 
immediately passes from the tube at the bottom of the furnace 
into the central shaft. The middle shaft has meanwhile been 
filled with these three mixtures in their proper order: — 

First: A mixture of sodium carbonate, carbon, sulphur and 

Second : Aluminium sulphate. 

Third : A flux, preferably a mixture of sodium and potassium 

This central shaft must be already strongly heated to com- 
mence the operation ; it is best to fill it with coke before charg- 
ing, and as soon as that is hot to put the charges in on the 
coke. Coke may also be mixed with the charges, but it is not 
necessary. The process then continues as follows : The water- 
gas enters the bottom of the shaft at a very high temperature. 
These highly heated gases, carbonic oxide and hydrogen, act 
upon the charges so that the first breaks up into a combina- 
tion of sodium sulphide and aluminium sulphide, from which, 
by double reaction with the second charge of aluminium sul- 
phate, free aluminium is produced. As the latter passes down 
the shaft, it is melted and the flux assists in collecting it, but is 
not absolutely necessary. Instead of producing this double 
sulphide, pure aluminium sulphide might be used for the first 
charge, or a mixture which would generate it ; or again pure 
sulphide of sodium, potassium, copper, or any other metallic 
sulphide which will produce the effect alone, in which case alu- 
minium is obtained alloyed with the metal of the sulphide. 
Instead of the first charge, a mixture of alumina, sulphur and 
carbon might be introduced. Or the aluminium sulphate of 
the second charge might be replaced by alumina. So one 


charge may be sulphide of sodium, potassium or any other 
metallic sulphide, and the second charge may be either alumina 
or aluminium sulphate. 

Messrs. Pearson, Turner, and Andrews* claim to produce 
aluminium by heating silicate of alumina or compound sili- 
cates of alumina and other bases with calcium fluoride and 
sodium or potassium carbonate or hydrate, or all of these to- 
gether. If other metals are added, alloys are obtained. 

G. A. Faurie,f of St. Denis, France, claims that by making 
an intimate mixture of alumina with carbon, moistening with 
sulphuric acid (thereby producing some sulphate) and in- 
tensely igniting the mass, aluminium is reduced, and globules 
of aluminium picked out of the resulting powder. Another 
metal may be added to produce an alloy. 

Reduction in Presence of or by Copper. 

Calvert and Johnson f obtained copper alloyed with alu- 
minium by recourse to a similar chemical reaction to that em- 
ployed to get their iron-aluminium alloy. Their mixture was 
composed of — 

20 equivalents of copper 640 parts. 

8 (24) " aluminium chloride 1076 " 

10 " lime 280 " 

"We mixed these substances intimately together, and after 
having subjected them to a high heat for one hour we found at 
the bottom of the crucible a melted mass covered with cuprous 
chloride, Cu^Clj, and in this mass small globules, which on 
analysis contained 8.47 per cent, aluminium, corresponding to 
the formula — 

5 equivalents of copper 160 9i'96 per cent. 

1 " " aluminium 14 8.04 " 


♦English Patent, 12332, Sept. 12, 1887. 
tU. S. Patent, July 22, 1890. 
X Phil. Mag. 1855, X. 242. 


" We made another mixture of aluminium chloride and cop- 
per in the same proportions as above, but left out the lime. 
We obtained an alloy in this case also, which contained 12.82 
per cent, aluminium, corresponding to the formula — 

3 equivalents of copper 96 87.27 per cent. 

I " "aluminium 14 12.73 " 

M. Evrard,* in order to make aluminium bronze, makes use 
of an aluminous pig-iron. (It is not stated how this alumin- 
ous pig-iron is made.) This is slowly heated to fusion, and 
copper is added to the melted mass. Aluminium, having more 
affinity for copper than for iron, abandons the latter and com- 
bines with the copper. After the entire mass has been well 
stirred, it is allowed to cool slowly so as to permit the bronze, 
which is heavier than iron, to find its way to the bottom of the 
crucible. M. Evrard makes silicon bronze in the same way by 
using siliceous iron. 

Benzont has patented the reduction of aluminium with cop- 
per, forming an aluminium-copper-alloy. He mixes copper, or 
oxidized copper, or cupric oxide, in the finest possible state, 
with fine, powdered, pure alumina and charcoal, preferably 
animal charcoal. The alumina and copper or copper oxide are 
mixed in equivalent proportions, but an excess of charcoal is 
used. The mixture is put in a crucible such as is used for 
melting cast-steel, which is lined inside with charcoal. The 
charge is covered with charcoal, and the crucible subjected first 
to a temperature near the melting point of copper, until the 
alumina is reduced, and then the heat is raised high enough to 
melt down the alloy. In this way can be obtained a succession 
of alloys, whose hardness and other qualities depend on the 
percentage of aluminium in them. In order to obtain alloys of 
a certain composition, it is best to produce first an alloy of the 
highest attainable content of aluminium, to analyze it, and then 

* Annales du Genie Civil, Mars, 1867, p. 189. 
t Eng. Pat. 1858, No. 2753. 


melt it with the required quantity of copper. The same pro- 
cess can be used for the reduction of alumina with iron or ferric 
oxide, only the carbon must in this case be in greater excess, 
and a stronger heat kept up longer must be used than when 
producing the copper-aluminium alloy. In contact with ferric 
oxide the alumina is more easily reduced than with the metallic 

Benzon further remarks that some of these alloys, as the 
ferro-aluminium, may be subsequently treated so as to separate 
out the metallic aluminium ; also that the iron alloy may be 
mixed with steel in the melting pot, or suitable proportions of 
alumina and carbon may be put into the melting pot. The iron 
alloy may be useful for many purposes, especially in the manu- 
facture of cast-steel. 

The question opened up by Benzon's statements is whether 
carbon reduces alumina in presence of copper. This has been 
the subject of many careful experiments, and the verdict of the 
most reliable observers is that at ordinary furnace temperatures 
it does not. This principle has been the subject of numerous 
patents, and before presenting the negative evidence on this 
point we will review the claims made in these patents. 

G. A. Faurie * states that he has succeeded in obtaining alu- 
minium bronze by taking two parts of pure, finely-powdered 
alumina, making it into a paste with one part of petroleum, 
and then adding one part of sulphuric acid. When the yellow 
color is uniform and the mass homogeneous, sulphur dioxide 
begins to escape. The paste is then wrapped up in paper and 
thrown into a crucible heated to full redness, where the petro- 
leum is decomposed. The calcined product is cooled, powdered 
and mixed with an equal weight of a metal in powder, e. g., 
copper. This mixture is put into a graphite crucible and 
heated to whiteness in a furnace supplied with blast. Amidst 
the black, metallic powder are found buttons of aluminium 
alloy. In an English patent,! by Mr. Faurie, it is further 

♦Comptes Rendus, 105, 494; Sept. 19, 1887. 
t English patent 10,043, -^"g- '^i J887. 


claimed that by making bricks out of the calcined alumina 
mixture and alloying metal, and using similar bricks of lixi- 
viated soda ashes mixed with tar for flux, the reduction can be 
effected in a cupola. 

Bolley,* at his laboratory in Zurich, and List.f at the royal 
foundry at Augsburg, have shown that by following the process 
claimed by Benzon the resulting copper contained either no 
aluminium, or at most a trace. 

In an experiment made by the author to test this point — 

40 grammes of copper oxide and copper, 
5 " alumina, 

5 " charcoal, 

were intimately mixed and finely powdered, put in a white-clay 
crucible and covered with cryolite. The whole was slowly 
heated to bright redness, and kept there for two hours. A 
bright button was found at the bottom of the crucible. This 
button was of the same specific gravity as pure copper, and a 
qualitative test showed no trace of aluminium in it. A friend 
of mine. Dr. Lisle, has repeated this experiment, taking the 
metal produced and returning it to another operation and re- 
peating this four times, but the resulting button scarcely showed 
a trace of aluminium. 

Dr. W. Hampe has also made an experimental test of this 
subject with the following conclusions: J — 

"The reduction of alumina by carbon, although often 
patented, is on thermo-chemical grounds highly improbable ; 
but since aluminium in alloying with copper, especially in the 
proportions 9.7 parts of the former to 90.3 parts of the latter 
(AlCu,), evolves much heat, it might be possible that the re- 

AI.O3 + 3C + 8Cu - 2AlCu^ + 3CO 

is exothermic. I therefore mixed alumina with the necessary 

* Schweizer Polytechnisches Zeitschrift, i860, p. 16. 

t Wagner's Jahresbericht, 1865, p, 23. 

I Chemiker Zeitung (Cothen), xii. p. 391 C1888). 


quantity of lamp-black and copper, in other cases evaporated 
together to dryness solutions of aluminium and copper nitrates, 
afterwards igniting them to oxides and adding the necessary 
amount of carbon. These mixtures were put into gas-carbon 
crucibles contained within plumbago pots with well-luted 
covers, and heated in a Deville blast-furnace to a temperature 
sufficient to frit together the quartz sand with which the space 
between the two crucibles had been filled. In no case was 
there a trace of aluminium produced, nor did the addition of 
any flux for the alumina affect the result in any way." 

The possibility of reducing aluminium sulphide by copper 
has been generally decided affirmatively. M. Comenge 
claimed that it was possible (see p. 411), Reichel * also stated 
unreservedly that copper fillings performed the reduction at a 
high temperature. In an experiment by the author, copper 
foil was used instead of copper filings, the latter not being im- 
mediately at hand, and the result was negative. As a similar 
test with iron filings gave a good result, it seems quite probable 
that copper would have performed the reduction under proper 

Andrew Mann,f of Twickenham, patents a process which 
may be stated briefly as follows : Aluminium sulphate is 
mixed with sodium chloride and heated until a reaction be- 
gins to take place. The mass is mixed intimately with lime, 
and to this mixture aluminium sulphate and ground coke 
added. This is calcined, the powder mixed with a metal; as 
copper, and melted down. In this case the slags are calcium 
sulphide and copper chloride, while aluminium bronze is ob- 

L. Q. Brin, of Paris,} claims to produce aluminium bronze by 
the following process : Sheet copper is cleaned by pickling 
and then covered with a mixture of 2 parts borax, 2 parts 
common salt, and i part sodium carbonate, made into paste 

♦ Journal fiir Pr. Chemie, xi. p. 55. 

t English Patent, 9313, June 30, 1887; German Patent, 457SS' J^ec. 20, 18S7. 

J English Patents, 3547-8-9, March 7, 1888; U. S. Patent, 410574, Sept. 10, 1889. 



with water. The metal is then put into a reverberatory fur- 
nace, heated to bright redness, and vapors of aluminium chlo- 
ride led over it, carried in by a current of inert gas. (It is 
stated that the vapors of aluminium chloride are produced by 
heating in a retort a mixture of clay, salt and fluorspar.) The 
aluminium compound is said to be decomposed, and the nas- 
cent aluminium to combine with the copper, forming ij4 to 2 
per cent, bronze at one operation, and by using this over it may 
be enriched to any extent desired. In a modification of this 
method, the coating put on the metal contains clay or other 
earth rich in alumina. It is also stated that the metal thus 
coated can be put into a cupola with alternate layers of fuel and 
run down to an alloy. 

Reduction by or in Presence of Iron. 

M. Comenge claims that aluminium sulphide is reduced by 
iron (see p. 41 1 ) ; the statement is repeated by a writer in the 
" Chemical News," i860; F. Lauterborn* states that the reduc- 
tion takes place at a red heat; Reichelj also records his success 
in this reaction ; finally, the author has obtained encouraging 
results.! I used a product containing 32.3 per cent, of alu- 
minium sulphide. On mixing this intimately with fine iron 
filings, and subjecting to a high heat for one and a half hours, 
the product was a loose powder in which were small buttons of 
metal. They were bright, yellower than iron, and contained by- 
analysis 9.66 per cent, of aluminium. 

H. Niewerth§ has patented the following process : " Ferro- 
silicon is mixed with aluminium fluoride in proper proportions 
and the mixture submitted to a suitable red or melting heat by 
which the charge is decomposed into volatile silicon fluoride 
(SiFi), iron and aluminium, the two latter forming an alloy. 
In order to obtain the valuable alloy of aluminium and copper 

* Dingier, 242, 70. 

t Jrnl. fur Pr. Chemie, xi. 55. 

I Graduating Thesis, Lehigh University, June, 1886. 

§ Sci. Am. Suppl., Nov. 17, 1883. 


from this iron aluminium alloy, the latter is melted with metallic 
copper, which will then by reason of greater affinity unite with 
the aluminium, while the iron will retain but an insignificant 
amount of it. On cooling the bath, the bronze and iron sepa- 
rate in such a manner that they can readily be kept apart. In 
place of pure aluminium fluoride, cryolite may advantageously 
be employed, or aluminium chloride may also be used, in which 
case silicon chloride volatilizes instead of the fluoride. Or 
again, pure silicon may be used with aluminium fluoride, cryo- 
lite, or aluminium chloride, in which case pure aluminium is 

Mr. W. P. Thompson* has taken out a patent in England| 
for the manufacture of aluminium and similar metals, which is 
carried out as follows : " The inventor employs as a reducing 
agent iron, either alone or conjointly with carbon or hydrogen. 
The operation is effected in an apparatus similar to a Bessemer 
converter, divided into two compartments. In one of these 
compartments is placed melted iron, or an alloy of iron, which 
is made to run into the second by turning the converter. This 
last compartment has two tuyeres, one of which serves to in- 
troduce hydrogen, while by the other is introduced either alu- 
minium chloride, fluoride, double chloride or double fluoride, 
with sodium, in liquid or gaseous state. In the presence of the 
hydrogen, the iron takes up chlorine or fluorine, chloride or 
fluoride of iron is disengaged, and aluminium mixed with car- 
bon remains as a residue. Then this mixture of iron, alumin- 
ium and carbon is returned to the other compartment, where 
the carbon is burnt out by means of a current of air. The 
mass being then returned to the chamber of reduction, the 
operation described is repeated. When almost all the iron has 
been consumed, the reduction is terminated by hydrogen alone. 
There is thus obtained an alloy of iron and aluminium. (The 
preparation of sodium does not require the intervention of 
hydrogen. A mixture of iron with an excess of carbon and 

* Bull, de la Soc. Chem. de Paris, 1880, xxiv. 719. 

t March 27, 1879, No. 2101. 


caustic soda (NaOH) is heated in the converter, when the so- 
dium distils off. When all the carbon has been burnt, the iron 
remaining as a residue may be converted into Bessemer steel. 
As iron forms an alloy with potassium, the method would 
scarcely serve for the production of that metal.) To obtain 
the pure aluminium, sodium is first prepared by the process in- 
dicated, the chloride or fluoride of aluminium is introduced into 
the apparatus in the other chamber, when the metal is reduced 
by the vapor of sodium. The chambers ought to be slightly 
inclined, and an agitator favors the reaction. The inventor 
intends to apply his process to the manufacture of magnesium, 
strontium, calcium and barium." 

That aluminium could be produced at all by the above re- 
actions is very improbable ; it is not likely that the reactions 
were even seriously tested. 

Calvert and Johnson * made experiments on the reduction 
of aluminium by iron, and the production thereby of iron- 
aluminium alloys. We give the report in their own words: — 

"We shall not describe all the fruitless efforts we made, but 
confine ourselves only to those which gave satisfactory results. 
The first alloy we obtained was by heating to a white heat for 
two hours the following mixture : — 

8 equivalents of aluminium chloride 1076 parts. 

40 " iron filings 1 120 " 

8 " lime 224 " 

"The lime was added to the mixture with the view of remov-. 
ing the chlorine from the aluminium chloride, so as to liberate 
the metal and form fusible calcium chloride, CaClj. Subtract- 
ing the lime from the above proportion, we ought to have 
obtained an alloy having the composition of i equivalent of 
aluminium to 5 equivalents of iron, or with 9.09 per cent, of 
aluminium. The alloy we obtained contained 12 per cent., 
which leads to the formula AlFe^. This alloy, it will be noticed, 
has an analogous composition to the one we made of iron and 

*Phil. Mag., 1855, A. 240. 


potassium, and like it was extremely hard, and rusted when 
exposed to a damp atmosphere. Still it could be forged and 
welded. We obtained a similar alloy by adding to the above 
mixture some very finely pulverized charcoal and subjecting it 
to a high heat in a forge furnace for two hours. This alloy 
gave on analysis 12.09 per cent.* But, in the mass of calcium 
chloride and carbon remaining in the crucible there was a large 
amount of globules varying in size from a pin-head to a pea, 
as white as silver and extremely hard, which did not rust in 
the air or in hyponitric fumes. Its analysis gave 24.55 P^"" 
cent, aluminium ; the formula AljFej would give 25 percent. 
Therefore this alloy has an analogous composition to alumina, 
iron ifeplacing oxygen. We treated these globules with weak 
sulphuric acid, which removed the iron and left the aluminium, 
the globules retaining their form, and the metal thus obtained 
had all the properties of the pure aluminium. 

"We have made trials with the following mixtures, but al- 
though they have yielded results, still they are not sufiSciently 
satisfactory to describe in this paper, which is the first of a 
series we intend publishing on alloys. This mixture was: — 

Kaolin 1 750 parts. 

Sodium chloride 1 200 " 

Iron 87s " 

"From this we obtained a metallic mass and a few globules 
which we have not yet analyzed." 

(See also Benzon's process, p. 414.) 

M. Chenot,t on the occasion of Deville's first paper on alu- 
minium being read to the French Academy, Feb. 6, 1854, an- 
nounced that in 1847, by reducing earthy oxides by means of 
metallic sponges, he had obtained a series of alloys containing 
up to 40 per cent, of the earth metals. He cited from a 

* In the original paper it is given as 1 2.09 per cent. iron. The inference is un- 
avoidable that this was a misprint, but it is not corrected in the Errata at the end of 
the volume. 

tComptes Rendus, xxxviii^4i5. 


memoir presented by him to the " Societe d'Encouragement," 
in 1849, in which he had said, "on taking precipitates of the 
earths, they are all reduced by the metallic sponge {e. g., that 
formed by reducing iron oxide in a current of carbonic oxide 
gas). In this manner I have made barides, silicides, aluminides, 
etc., all of which are beautiful silver-white, very hard and un- 
oxidizable in air or in contact with acid vapors. They are 
fusible, can be cast, and work perfectly under the hammer." 

Faraday and Stodart* made an exhaustive investigation on 
the preparation of iron-aluminium alloys, being started on this 
line by finding that Bombay " wootz" steel contained O.O128 to 
0.0695 psi" cent, of aluminium, while no metals of the earths 
were to be found in the best English steels. This led to the 
conclusion that the peculiar properties of the former, especially 
the " damasceening," were due to the small amount of alumin- 
ium. These scientists commenced by taking pure steel or 
sometimes good soft iron and intensely heating it for a long 
time imbedded in charcoal powder. Carbides were thus 
formed, having a very dark gray color, and highly crystalline. 
Average analysis of this product gave 5.64 per cent, carbon. 
This was broken and powdered in a mortar, mixed intimately 
with pure alumina, and heated in a closed crucible for a long 
time at a high temperature. An alloy was obtained of a white 
color, close granular texture and very brittle, containing 3.41 
per cent, of aluminium, with some carbon. 

When 40 parts of this alloy were melted with 700 parts of 
good steel (introducing O.184 per cent, of aluminium) a malle- 
able button was obtained which gave a beautiful damask on 
treatment with acids ; while 6"] parts of the alloy with 500 of 
steel (introducing 0.4 per cent of aluminium) gave a product 
which forged well, gave the damask and " had all the appreci- 
able characters of the best Bombay wootz." This appears to 
be very strong synthetic evidence that alumina is reduced to a 
small extent even in the rude hearths in which the Indian steel 
is manufactured. Karsten, however, could not find weighable 

* Quarterly Journal, ix. 320. 


quantities of aluminium in specimens of wootz, nor could 
Henry, a very expert analyst. The latter suggested that Fara- 
day was misled by the alumina contained in intermingled slag, 
yet the latter obtained alumina without silica in his analyses. 

In the light of more recent developments we would accept 
Faraday's results as being very near the truth in the matter. 

Ledebuhr* quotes an analysis made by Griiner in which 0.50 
per cent, of aluminium was found in cast-iron containing be- 
sides 2.30 per cent, of carbon and 2.26 per cent, of silicon. 
This would tend to show that under certain conditions iron 
takes up aluminium in the blast-furnace. Karsten, however, in 
his many analyses of malleable iron, steel and cast-iron, only 
found aluminium in unweighable quantities. Griiner and Lauf 
stated that aluminium is reduced in small quantities in the blast 
furnace if the temperature is high and the slag basic; a large 
addition of lime thus increases the reduction of alumina and 
hinders that of silica. Most pig irons contain very small 
amounts of aluminium, but some English varieties contain 0.5 to 
i.o per cent, and several Swedish pig irons 0.75 per cent. 
Schafhautl:j: found as much as i.oi per cent of aluminium in a 
gray iron, and was led to consider silicide of iron and aluminide 
of iron as characteristic components of gray iron. Lohage§ 
states that adding alumina in the manufacture of cast-steel has 
a great influence on the grain and lustre of the steel, the effect 
being doubtless due to a minute quantity of aluminium taken 
up. Silicates of magnesium and aluminium are formed at the 
same time, and separating out, float on the surface of the molten 
steel. Corbinjl reports 2.38 per cent, of aluminium in chrome 
steel ; but Blair,ir of Philadelphia, found no more aluminium in 
chrome than in other steels. This chemist has examined many 

* Handbuch der Eisenhiittenkunde, p. 265. 

t Berg u. Hiittenmannische Zeitung, 1862, p. 254. 

X Erdman's Journal fr. Pr. Chemie, Ixvii. 257. 

§ Berg u. Huttenmannische Zeitung, 1861, p. 160. 

II Silliman's Journal, 1869, p. 348. 

t H. M. Howe, E. and M. J., Oct. 29, 1887. 


irons and steels particularly for aluminium, and reports that 
nearly always it exists as such in steel, but never more than a 
few thousandths of i per cent., say 0.032 per cent as a maxi- 
mum. He has further been unable to connect its presence 
with any peculiarity in the properties of the metal or its mode 
of manufacture. 

G. H. Bilhngs,* of the Norway Iron Works, Boston, made 
the following experiment on reducing alumina in contact with 
iron : — A soft iron was used containing a trace of sulphur and 
phosphorus, no manganese and only 0.08 per cent, of carbon. 
The mixture was made of 

12 parts emery. 
18 " alumina. 
I " pulverized charcoal. 
36 " fine iron turnings. 

These were mixed thoroughly, and heated to whiteness for 48 
hours. The melting resulting showed a solid, homogeneous 
fracture with a fine crystalline structure resembling steel with i 
per cent, of carbon, and contained on analysis 

0.20 per cent, of carbon. 
0.50 " aluminium. 

It was also found that if this quantity of aluminium was added 
to a pot of molten iron, the product obtained exhibited the same 
characteristics as the above. 

Another attempt to produce iron-aluminium alloys directly 
is stated in E. Cleaver's patent specifications as follows : f 
Four parts of aluminium sulphate in solution are mixed with 
one part of lamp-black, the mixture dried and heated to the 
highest temperature attainable by using coal-gas and oxygen 
in a lime-lined furnace similar to those used for melting plati- 
num. Excess of reducing gas is maintained. The charge is 
cooled in the furnace, removed, mixed with twenty-times its 
weight of finely-divided cast-iron, and fused in a steel melting 

* Transactions American Inst. Mining Engineers, 1877, p. 452. 
t English Patent, 1276, Jan. 26, 1887. 


furnace. If copper is used, a bronze results. The alloying 
metal may be added in the gas furnace, but this is not recom- 
mended as economical. This inventor also claims that alu- 
minium ferrocyanide, either alone or with carbon, can be 
decomposed in the above-described gas furnace, yielding a rich 
iron-aluminium alloy. As a higher heat than before is needed, 
it is recommended that the oxygen be previously heated. The 
principal difficulty in this latter process would apparently be to 
procure the aluminium ferrocyanide to operate on. 

Mr. Ostberg,* connected with the Mitis process for making 
wrought-iron castings, stated that the ferro-aluminium used in 
that process in Sweden was made by adding clays to iron in 
process of smelting, that it contained 7 to 8 per cent, of alu- 
minium, and could be made very cheaply. Inquiries made for 
further particulars about this process have received no satisfac- 
tory reply, and there is no outside confirmation of the above 
statement to be found. 

Brin Bros, claim that they can alloy aluminium in small 
quantities with iron (see p. 417). Besides the processes de- 
scribed as most suitable for producing bronze, they also state 
that if soft strap-iron is coated with the flux composed of clay 
and salt and heated to over 1000° C. in a mufHe or a blowpipe 
flame, the iron absorbs aluminium and becomes tough and 
springy, having many of the properties of steel. They also 
claim that by simply charging broken lumps of cast-iron into 
a cupola with alternate layers of common clay and a flux, the 
metal run down contains as much as 1.75 per cent, of alumin- 
ium, yielding a very fluid, strong iron, which runs into the 
thinnest castings. The London papers state that the alloys 
thus produced assuredly contain aluminium, and that the con- 
tained aluminium does not cost over 25 cents per lb. 

A newspaper report speaks of exactly similar processes be- 
ing operated by an aluminium company in Kentucky. (See 
also p. 403.) It is said that they charge a cupola with scrap- 
iron, pig-iron, coke, clay, and a flux,- and that on melting the 
* Engineering and Mining Journal, May 15, 1886. 


charge down and pouring, the castings produced are similar to 
the best steel, the fracture of the metal being white, slightly- 
fibrous, and free from blow-holes. It is stated, further, that the 
castings, on analysis, contained 1.7 per cent, of aluminium. 
Scrap-iron is also treated in the same way as reported by Brin 
Bros., being simply coated with a pasty mixture of clay and a 
flux and heated almost white-hot, when the iron absorbs alu- 

The Aluminium Process Company of Washington, D. C, 
own several patents granted to W. A. Baldwin, of Chicago, 111. 
In one of these,* a bath is formed by fusing together 4 parts of 
ground clay, 12 parts of common salt and i part of charcoal 
powder. The metal to be alloyed, e. g., an iron bar, is thrust 
into the bath, which is not hot enough to melt it, and allowed 
to remain some time, with occasional stirring, until the alloying 
is complete. In the case of metals with low fusing points the 
metal may be melted with the mixture. It is claimed that the 
metal takes up a small percentage of aluminium. With a more 
highly aluminous material than clay, the proportions of salt and 
carbon are to be increased proportionately. In a modification 
of this process adopted for foundry practice,! a mixture of clay, 
salt and carbon, similar to the above, is put into a large ladle 
and the molten iron tapped directly from the cupola on to it. 
A brisk stirring up of the iron takes place, much scum rises to 
the surface, and the resulting iron is more fluid, can be carried 
further before setting, and makes sounder and stronger castings 
than similar iron not treated. The resulting iron does not 
contain enough aluminium to be detected by quantitative 
analysis. Old-fashioned fluxes used long ago in foundries were 
similar in composition to this mixture used by Baldwin, and 
were found efficacious in freeing dirty iron from slag and other 
impurities. It is hard to see how this latter process is anything 
but the use of a common flux, with a different explanation as 
to how it acts — the explanation being probably the most ques- 

* English Patent, 2584, Feb. 21, 1888. 
tU. S. Patent, 380161, March 27, 1888. 


tionable part of the whole. The first-mentioned process, how- 
ever, has the merit of novelty, and pieces of poor iron treated 
by it are made springy and much like steel, but whether this is 
due to absorption of aluminium is doubtful. 

The Williams Aluminium Company, of New York City 
(works at Newark, N. J.), manufacture an alloy which they call 
aluminium-ferro-silicon and sell for foundry use. When first 
introduced, this alloy was represented to contain 10 per cent, of 
aluminium, but several analyses disproved this, and afterwards 
the alloy was sold simply on the guarantee of what it would 
accomplish. The metal since sold by this company does 
contain a small amount of aluminium, and its action on poor 
foundry iron is similar to that of other brands of ferro-alumin- 
ium ; therefore, as long as no certain percentage of aluminium 
is now claimed, the company is certainly doing a legitimate 
business. (For method of using, etc., see Chap. XVI.) The 
alloy is made by melting down a mixture of iron filings, clay, 
salt,' charcoal, and another flux whose composition is not 
divulged. This is put into cast-iron pots and the whole charge, 
crucibles and all, run down in a furnace of peculiar design 
constructed by Mr. Williams. The capacity of the plant is 
about 1000 lbs. of alloy a day, which is broken by small stamps 
into pieces of about an inch diameter and sold at 10 cents 
per lb. Mr. Williams also experimented on manufactur- 
ing aluminium bronze by the same methods, and a sample 
piece forwarded the author contained a small percentage of 

G. Bamberg, London,* claims to produce alloys of alumin- 
ium by taking iron or copper rich in silicon, putting them 
molten into a Bessemer converter, raising the temperature by 
blowing air through in the usual way, and then passing in 
vapors of aluminium chloride or aluminium-sodium double 

Mr. Chas. T. Holbrook, connected with the well-known 

* English Patent, 7666, May 8, 1889. 


iron-masters Marshall Brothers, of Philadelphia, has reported 
the existence of aluminium in the pig-iron made at the Marshall 
furnace, Newport, Perry county. Pa. Mr. Holbrook has made 
many tests on the efTect of adding aluminium to cast-iron, and 
on one occasion, happening to see pig-iron being tapped from 
this furnace, he remarked that it flowed as if it contained alu- 
minium, judging by its bright, silvery appearance. An analy- 
sis of the pig-iron was at once made, and confirmed this opin- 
ion by showing 0.267 per cent. Since then, numerous analyses 
have shown an average of 0.20 to 0.50 per cent., while for a 
short period, when the furnace was working much hotter than 
usual, over one per cent, has been obtained. The question as 
to what circumstances favor this considerable reduction of alu- 
mina is probably to be referred to the nature of the combina- 
tion in which it occurs in the ores. It is found, for instance, 
that when using ores from the Juniata Valley, the iron was 
richer in aluminium than when a larger amount of ore from a 
distance was used. The composition of the local ores is there- 
fore of interest, and the following are copied from a letter of 
Mr. Holbrook's : * 

Grafton Fossil Ore. Juniata Hematite 
(Dried at ioo° C.) (Dried at ioo° C.) 

Silica 10.680 17.580 

Alumina 12.469 5-692 

Ferric oxide 73.481 6l-l33 

Manganese oxide 0.431 0.700 

Fhosphoric acid i .1 70 0.895 

Sulphuric acid none none 

Lime trace trace 

Magnesia none none 

Combined water and organic matter 2.700 10.700 

100.931 96.700 

The pig-iron made contains 3 to 5 per cent of silicon, and is 
a first-class foundry iron. 

Most aluminium was reduced when using the Grafton ore, 
and it appears in this ore the alumina is present in a much 

* Iron Age, June II, 1891. 


larger proportion than it could be if combined with silica alone. 
In other words, if the siKca were all present as ordinary clay, 
it could not be combined with over 9 per cent, of alumina. It 
is to me a very probable conclusion that the ore either con- 
tains free alumina, or possibly, some Hercynite, the aluminate of 
iron, FeO.AljOs. This compound would evidently reduce to an 
iron-aluminium alloy. Whatever be the real reason for it, the 
fact remains that this is a real instance of aluminium being re- 
duced from alumina in the presence of iron, at the temperature 
of a blast furnace of moderate size. This temperature prob- 
ably never exceeds 2000° C. 

Pearson and Pratt* have patented the following means for 
facilitating the reduction of aluminium in the iron blast-fur- 
nace. Calcined iron ore is mixed with a considerable quantity 
of clay and the necessary amount of coke, and charged into a 
blast-furnace with fluorspar for a flux instead of limestone. 
Cast-iron containing aluminium results. 

Reduction by or in Presence of Zinc. 

M. Beketoff,! was not able to reduce vapor of aluminium 
chloride by vapor of zinc, although silicon chloride under the 
same conditions was readily reduced. 

M. Dullo| observes that the double chloride of aluminium 
and sodium, which he makes directly from clay, may be re- 
duced by zinc. He says, " The reduction by zinc presents no 
difificulties, but it is less easy than with sodium. An excess 
of zinc should be employed, which may be got rid of after- 
wards by distillation. The, metal thus prepared possesses all 
the characteristics and all the properties of that obtained from 
bauxite with sodium." 

M. N. Basset,§ a chemist in Paris, patented a somewhat sim- 

* English Patent, 18048, Nov. 12, 1889. 
t Bulletin de la Societe Chemique, 1857, p. 22. 
J Bull, de la Soc. Chem., i860, v, 472. 
§Le Genie Industriel, 1862, p. 152. 


ilar process for obtaining aluminium. If the statements are 
correct they are of great value. The paper is as follows : 

" All the metalloids and the metals which form by double de- 
composition proto-chlorides or sesqui-chlorides more fusible or 
more soluble than aluminium chloride may reduce it or even 
aluminium-sodium chloride. Thus, arsenic, bismuth, copper, 
zinc, antimony, mercury, or even tin or amalgam of zinc, tin, 
or antimony may be employed to reduce the single or double 
chloride. The author employs zinc in preference to the others 
in consequence of its low price, the facility of its employment, 
its volatility, and the property which it has of metallizing easily 
the aluminium as it is set free. When metallic zinc is put in 
the presence of aluminium-sodium chloride, at 250° C. to 300° 
C, zinc chloride, ZnCl^, is formed and aluminium is set free. 
This dissolves in the zinc present in excess, the zinc chloride 
combines with the sodium chloride, and the mass becomes lit- 
tle by little pasty, then solid, while the alloy remains fluid. If 
the heat is now raised, the mass melts anew, the zinc reduces a 
new portion of the double chloride and the excess of zinc en- 
riches itself in aluminium proportionately. These facts consti- 
tute the basis of the following general process : One equiva- 
lent of aluminium chloride is melted, two of sodium chloride 
added, and when the vapors of hydrochloric acid are dissipated, 
four equivalents of zinc, in powder or grain, are introduced. 
The zinc melts rapidly, and by agitation the mass of chloride 
thickens and solidifies. The mass is now composed of the 
chlorides of aluminium, zinc and sodium, and remains in a 
pasty condition on top of the fluid zinc containing aluminium. 
This pasty mass is removed, piled up in a crucible or in a fur- 
nace, and bars of the fluid alloy of zinc and aluminium ob- 
tained from a previous operation are placed on top of it. This 
is gradually heated to bright redness, and kept there for an 
hour. The melted mass is then stirred with a rake and poured 
out. It is an alloy of the two metals in pretty nearly equal 
proportions. This alloy, melted with some chloride from the 
first operation, furnishes aluminium containing only a small 


per cent, of zinc, which disappears by a new fusion under alu- 
minium chloride mixed with a little fluoride, providing the 
temperature is raised to a white heat and maintained till the 
cessation of the vapors of zinc, air being excluded. 

" The metal is pure if the zinc employed contained no foreign 
materials or metals. In case the zinc contains iron, or even if 
the aluminium chloride contains some, the metallic product of 
the second operation may be treated with dilute sulphuric acid 
to remove it. The insoluble residue is washed and melted 
layer by layer with fluorspar or cryolite and a small quantity of 
aluminium-sodium chloride, intended solely to help the fusion." 

Mr. Wedding,* makes the following remarks on this process : 

" It is some time since Mr. Basset established the possibility 
of replacing sodium by zinc in the manufacture of aluminium. 
Operating on aluminium-sodium chloride with granulated zinc, 
the reduction takes place towards 300°. The reduced alumin- 
ium dissolves in the excess of zinc, while the zinc chloride 
formed combines with the sodium chloride, forming a pasty 
mass if the heat is not raised. Under the action of heat the 
alloy enriches itself in aluminium, because the zinc volatilizes. 
The zinc retained by this alloy is completely eliminated by fus- 
ion with double chloride and a little fluorspar. The tempera- 
ture ought' to be pushed at least to a white heat, and main- 
tained till no vapor of zinc escapes, air being excluded during 
the operation. These results I have confirmed, having sub- 
mitted the experiments of Mr. Basset to an attentive examina- 
tion, and I recommend its use. However, the process de- 
mands very much precaution, because of the high temperature 
which it necessitates. Another chemist, Mr. Specht, even in 
i860 decomposed aluminium chloride by zinc, and has the 
same report to make — that he thinks the process will be some- 
time advantageously practised on a large scale." 

The author made the following experiment to determine if 
cryolite would be reduced by zinc : One pound of finely pow- 

* Journal de Pharm. [4] iii. p. 155 (1866.) 


dered cryolite was melted in a graphite crucible, and 6 ounces 
of granulated zinc dropped into it. No perceptible reaction 
took place except the volatilization of zinc when the crucible 
was uncovered, and the metal obtained after 15 minutes' treat- 
ment contained on analysis 0.6 per cent, of aluminium. 

Mr. Fred J. Seymour * patented the reduction of aluminium 
by zinc, making the following claim : An improvement in ex- 
tracting aluminium from aluminous earths and ores by mixing 
them with an ore of zinc, carboniferous material and a flux, 
and subjecting the mixture to heat in a closed resort, whereby 
the zinc is liberated, is caused to assist in bringing or casting 
down the aluminium in a metallic state, and the alloy of alu- 
minium and zinc is obtained. 

A furnace was put up in the early part of 1884, somewhere 
in the vicinity of Cleveland, in a description of which by a 
newspaper correspondent we are told that steel retorts were 
charged with a mixture of zinc ore 100 parts, kaolin 50, carbon 
(either anthracite coal or its equivalent of some hydrocarbon) 
125, pearlash 15, common salt 10: the heat necessary being 
about 1400° C. In a second patent, f Mr. Seymour claimed 
that by heating the same mixture in a retort and introducing 
air he volatilized oxides of aluminium and zinc, which were 
caught in a condenser, mixed with carbon, and reduced in a 
crucible. Immediately after the issue of this patent, the Amer- 
ican Aluminium Company was organized in Detroit with a cap*- 
ital stock of $2,500,000 to operate the patents of a Dr. Smith, 
under whose name processes similar to the above had been 
patented in Great Britain and France. A works was then 
started at Findlay, Ohio, using natural gas for fuel. A gentle- 
man who saw the plant described it as a reverberatory furnace 
into which a charge of 800 lbs. of zinc ore, 900 of native alum, 
and 300 of charcoal was put. On heating very strongly by gas 
with plenty of air admitted, zinc oxide was volatilized (Mr. Sey- 

*U. S. Pat., No. 291631, Jan. 8, 1884. 
tU: S. Patent, 337,996, March 16, 1886. 


mour claimed that it carried alumina with it), and was con- 
densed by passing the gases through large copper condensers. 
This fume was then collected, mixed with carbon and a metal, 
and run down in a crucible to an alloy. It was clairiied that 
the plant had a capacity of 600 lbs. of pure aluminium a day, 
and had presumably been in operation over a year, yet there 
was not over 20 lbs. of alloys or j^ lb. of pure aluminium to 
show for it. On closer inspection so plain indications of fraud 
were visible in the last part of the operation that, although ver- 
itable aluminium alloys were taken out of the crucible, the gen- 
tlemen referred to refused to believe that the aluminium was 
produced in the process. As the sequel to this it can be 
stated that in the middle of 1889 the executive committee for 
the stockholders, being fully satisfied of the worthlessness of 
the process, called for a meeting to wind up the company. 
One month later Mr. Seymour died. The story went the 
rounds of the daily press that the one metallurgist who com- 
manded the secret of obtaining cheap aluminium had died, tak- 
ing the talisman with him, and a vivid picture was drawn of the 
manner in which the secret had been preserved by means of 
twelve-foot palisades, double-bolted doors, and by working at 
the midnight hour. Alas, the secret was out one month before 
he died ! 

A method of reducing cyanide of aluminium by means of 
zinc is the subject of a patent granted F. Lauterborn.* Four- 
teen parts of aluminium sulphate dissolved in twice its weight 
of water is precipitated by thirteen parts of ferro-cyanide of 
potassium dissolved in four times its weight of water ; the pre- 
cipitate of ferro-cyanide of aluminium is collected and dried. 
This substance is then mixed with slightly less than one-half its 
weight of dry, anhydrous sodium carbonate, put in a crucible 
and ignited with as little admission of air as possible. The 
ferro-cryanide is decomposed, iron carbide separates out, and 
besides sodium cyanate, the double cyanide of aluminium and 

* German Patent, 39915 (1887), 


sodium (Al2Cy6+3NaCy) is obtained as a melted mass which 
is poured away from the heavier iron carbide at the bottom of 
the crucible. If two parts of this salt are then ignited with one 
part of zinc in a covered crucible, aluminium separates as a 
regulus, while double cyanide of sodium and zinc remains as 
slag. The slag is dissolved in water and treated with metallic 
iron, whereby metallic zinc is precipitated out and a solution of 
ferro-cyanide of soda remains and can be used over in the pro- 

I do not know whether Lauterborn has ever succeeded in 
carrying out this process ; it would appear at the very begin- 
ning that the precipitation of aluminium ferro-cyanide, although 
appearing a priori possible, has always been found impractic- 
able, alumina being precipitated. 

J. Clark, of Birmingham, England, has taken out several pat- 
ents in England and one in Germany. In the first,* hydrated 
aluminium chloride is to be mixed with lime, iron, zinc, am- 
monia, or any other substance which combines readily with 
chlorine, and finely divided coke. After drying the mixture it 
is introduced into the iron blast furnace or blown into the Bes- 
semer converter, an iron-aluminium alloy being thus produced. 
In a second patent,! hydrated aluminium chloride is to be 
mixed with 2 ^2 parts of granulated zinc and i part of iron 
turnings or borings, or the alloy of the zinc and iron known as 
"zinc dross" might be used. The mass is let stand 24 hours 
and dried. The orange-colored powder resulting is mixed 
with borax (!) or any suitable flux, and put in a crucible with 
20 parts of fine granulated copper and melted down. After 
about an hour the zinc and iron present have probably volatil- 
ized as chlorides, while aluminium bronze remains. When the 
copper is to be alloyed in large quantities, it may be melted on 
the hearth of a reverberatory furnace and the prepared powder 
stirred into it. 

* English Patent, 15946, Dec. 6, 1886. 

t English Patent, 10594, Aug. 18, 1886; German Patent, 40205, (1887). 


Dr. J. D. Lisle, of Springfield, Ohio, made the following ex- 
periments in reducing aluminium sulphide by zinc. The sul- 
phide was made in a crucible, mounted on trunnions to keep 
the contents agitated, in which alumina was placed and treated 
with superheated carbon bi-sulphide vapor. While the con- 
tents were still hot, solid or melted zinc was put into it, the 
heating continued, the crucible constantly agitated and a cur- 
rent of hydrocarbon gas led into the crucible through a hole in 
the cover. The outcoming gas contained some sulphuretted 
hydrogen, seeming to show that reduction was accomplished 
partially, at least, by the hydrocarbon. After heating about 
half an hour the openings were sealed up and the apparatus 
cooled. The product was analyzed, and the result of nine ex- 
periments gave results varying from 1.06 to 10 per cent, of 
aluminium present in the zinc. The richest alloy was distilled 
carefully, leaving aluminium which analyzed 98 per cent, pure 
and contained only a trace of zinc. It is difficult, however, to 
see any possibility of a commerical process being worked up 
on these lines which could compete with the present methods 
of extraction. 

G. Bamberg, of London,* claims that if metallic zinc is vapor- 
ized it will reduce the vapor of aluminium chloride. He puts 
metallic zinc in a retort heated sufficiently to distil off zinc 
vapor, and in another retort is placed aluminium chloride or 
aluminium-sodium chloride. The vapors mix in a third highly 
heated chamber, in which they react, and there is condensed an 
alloy rich in aluminium. This is heated to whiteness in an- 
other retort to distil off the zinc. 

Reduction by Lead. 

According to the invention of Mr. A. E. Wilde, f of Notting 
Hill, lead or sulphide of lead, or a mixture of the two, is melted, 
and in a molten state poured upon dried or burnt alum. The 

* English Patent, 7667, May 8, 1889. 
t Sci. Am. Suppl., Aug. 11, 1887. 


crucible in which the mass is contained is then placed in a fur- 
nace and heated, with suitable fluxes. The metal, when poured 
out of the crucible, will be found to contain aluminium. The 
aluminium and lead can be subsequently separated from each 
other by any known means, or the alloy or mixture of the two 
metals can be employed for the various useful purposes for 
which lead alone is more or less unsuited. 

Reduction by Manganese. 

Walter Weldon * claimed to melt together cryolite with cal- 
cium chloride or some other non-metallic chloride or sulphide, 
and then to reduce the aluminium chloride or sulphide pro- 
duced by manganese, also adding metallic sodium to promote 
the reaction. Of course, the manganese would be used as ferro- 
manganese or spiegeleisen, and an iron alloy produced. Dr. 
Green, of Philadelphia, mixed powdered spiegeleisen with cryo- 
lite, placed the mixture in a graphite crucible, and heated it 
close to the ports of an open-hearth steel furnace until it soft- 
ened, yet the iron contained afterwards only 0.3 per cent, of 

When we examine closely the heat of combination of man- 
ganese and aluminium compounds (Chapter VIII.) we see that 
in the combinations with chlorine and sulphur manganese is 
the stronger element, and we should therefore be led to expect 
that manganese would reduce aluminium chloride or sulphide. 
On the other hand, in the oxides, and particularly in the flu- 
orides, aluminium is the stronger element. We should there- 
fore expect aluminium to reduce manganese oxide and fluoride. 
This is, in fact, just what occurs. Drs. Green and Wahl, of 
Philadelphia, are, in fact, producing pure manganese on a com- 
mercial scale by dissolving manganese oxide in melted cryolite 
and adding metallic aluminium, f 

* English Patent, No. 97 (1883). 

t See Journal of the Franklin Institute, March, 1893, P- 218. 

reduction of aluminium compounds. 437 

Reduction by Magnesium. 

Magnesium develops more heat in forming compounds than 
aluminium does (see p. 229), which would indicate that it 
would reduce aluminium compounds easily. Only one or two 
statements on this point can be found. Margottet * states that 
magnesium will decompose molten cryolite, setting the alu- 
minium at liberty. R. Gratzel f patents the reduction of a 
double fluoride of aluminium and potassium or sodium by 
metallic magnesium, or by conducting magnesium vapor into 
the liquid compound. 

Roussin I states that magnesium does not precipitate alu- 
minium in a metallic state from its solutions. To test this 
point I placed a strip of magnesium in solution of aluminium 
sulphate, when magnesium sulphate went into solution and a 
precipitate of alumina was formed. It is apparent that the 
aluminium is first precipitated in the metallic state and promptly 
oxidized by the water as fast as set free, in a manner strictly 
analogous to the production of alumina at the negative pole 
when electrolyzing aqueous aluminium solutions. 

In a patent awarded Count R. de Montgelas, of Philadel- 
phia,§ it is stated that aluminium chloride is mixed with litharge, 
charcoal and common salt ; fused and crushed. It is then re- 
melted with magnesium filings and potassium chloride. After 
cooling it is again crushed, mixed with more potassium chloride 
and nitrate of potash, fused, poured into water, and the globules 
of aluminium separated out. 

Clemens Winkler || states that when alumina is heated with 
powdered magnesium, in a current of hydrogen gas, it gives a 
nearly black powder containing over 40 per cent, of a sub-oxide 
of aluminium to which he assigns the formula AlO, together 

* Fremy's Ency. Chim. 
t English Patent, 14325, Nov. 25, 1885. 
{ Jrnl. de Pharm. et de Chimie, iii., 413. 
§ English Patent, 10606, August 18, 1886. 
II Berliner Berichte, xxiii. p. 772. 


with magnesium oxide and unaltered alumina. The last two 
are partly combined to magnesium aluminate. No metallic 
aluminium was obtained. 

Reduction by Antimony. 

F. Lauterborn * proposes to decompose aluminium sulphide 
by means of antimony and carbon. One hundred parts of 
dried aluminium sulphate is mixed with 50 parts of charcoal 
and 72 parts of metallic antimony ; some sodium carbonate and 
fluorspar is then added, and the mixture melted. It is claimed 
that antimony sulphide and aluminium are found in the pro- 
duct, the former being in the bottom of the crucible. In a 
modification of this process it is claimed t that if a mixture of 
dried aluminium sulphate, sodium carbonate, and antimony 
sulphide (stibnite) is put into a shaft filled with incandescent 
coke, antimony will be first set free by the reaction 


and that these products act further on the aluminium sulphate, 
setting free aluminium by the reaction 

2 Al, (SO*) 3+ 6Na,S+4Sb+ 1 2C=4Na2SbS3f4Al-|- 1 2CO,. 

The sodium sulph-antimonide can be smelted over with soda 
and antimony regained. 

These extraordinary formulas have little or no basis in chem- 
ical science. Dr. Fischer J says plainly that they are false. 

The author tried by direct experiment to reduce aluminium 
sulphide by antimony, fusing down a mixture of powdered 
antimony with aluminium sulphide, but the button of metal 
obtained did not contain a trace of aluminium. 

* German Patent, 32126(1885). 

t Dingier, 256, 226 and 233. 

I Wagner's Jahresbericht, 1885. 

reduction of aluminium compounds. 439 

Reduction by Tin. 

J. S. Howard and F. M. Hill, assignors to the Aluminium 
Product Company, of New York, make the following statements 
in their patent specifications:* "Some aluminous material is 
boiled in muriatic acid, cooled, mixed with Spianish white or 
lime, the free acid evaporated off at a high temperature, and 
the heat finally increased to about 1000° F., thereby volatiliz- 
ing any iron present as chloride. The product thus obtained 
is put into a lime-lined crucible along with lime, charcoal, fluor- 
spar, cryolite and potassium bisulphate, the whole covered with 
chloride of tin, and finally by common salt. This is subjected 
to a smelting temperature, when on cooling it is claimed that 
an alloy of tin and aluminium is obtained. To separate the 
aluminium the alloy is melted with lead or bismuth, which alloy 
with the tin, letting the aluminium float on the surface along 
with oxides and other impurities. This is skimmed off and 
purified by exposure to heat on a bed of porous material." 

While we cannot altogether deny the possibility of metallic 
tin reducing aluminium chloride, yet it is not at all probable 
that this salt remains after the first ignition. It is also improb- 
able that tin would reduce cryolite in the latter operation, but 
there is no direct evidence to contradict the above statement. 

In an experiment made by the author, aluminium sulphide 
was heated with tinfoil for a short time at a red heat. The 
resulting metal contained 0.52 per cent, of aluminium, and from 
the proportions of tin and sulphide used, it was evident that a 
large part of the aluminium sulphide present had been reduced. 

Dr. J. D. Lisle, of Springfield, Ohio, has made experiments 
of the same nature on a more extensive scale. Aluminium 
sulphide was put in a crucible placed on trunnions, so as to 
permit agitating the contents, and metallic tin in flakes was put 
in with it. The crucible was closed, and when at a bright-red 
heat a stream of super-heated gasoline vapor led into the cru- 

* U. S. Patent, 378,136, Feb. 21, 1888. 


cible. Hydrogen sulphide was copiously formed, but an ex- 
plosion of gasoline vapor terminated the experiment. The 
crucible was not broken, and when cool was opened. It con- 
tained some shot-metal, which was collected and melted into a 
button. This was the color of steel, very hard and tough, and 
upon analysis gave 19.32 per cent, of aluminium. In another 
experiment equal parts of aluminium sulphide and tin, in flakes, 
were mixed together, placed in a graphite crucible, covered 
with powdered charcoal, and kept at a red-heat for one-half 
hour. The resulting button contained 9.8 per cent, of alu- 

While such experiments are interesting as showing the pos- 
sibility of tin reducing an aluminium compound, yet pure 
aluminium cannot be made from the tin alloy, and metallic tin 
is so costly a reducing agent, being worth about 1 5 cents per 
pound, that this means of reduction can never be commercially 
applied, since aluminium itself is selling so cheaply. 

Reduction by Phosphorus. 

L. Grabau * patented the following process, in 1883, but has 
since virtually admitted that this method was not successful. 
The proposition was as follows : A rich alloy of aluminium 
and phosphorus is made by melting the two elements together 
or by fusing aluminium with phosphor salts and a reducing 
agent. The alloy is crushed, mixed with alumina or clay or 
aluminous fluorides, covered with coal dust and heated to in- 
candescence in a crucible. It was claimed that the phosphorus 
combined with the oxygen or fluorine of the aluminous com- 
pound, the metal produced uniting with the aluminium already 
present. To produce any given alloy, the phosphide of the 
required metal is substituted for the aluminium-phosphorus 
alloy. Mr. Grabau broadens out his specifications into more 
probable fields by adding that " manganesic or carburetted 
metals may be used instead of the phosphide alloys, as reducing 

* English Patent, 5798, Dec. 18, 1883. 


agents." It is almost needless to say that the generally small 
heat of combination of phosphorus compounds would show 
that the reactions proposed are in a high degree improbable. 

Reduction by Silicon. 

M. Wanner * makes the following general claims : The pro- 
duction of aluminium by treating a fused bath of aluminium 
fluoride or aluminous fluorides, while in a molten metallic bath 
and protected from oxidizing agents, with a reagent whose 
elements dissociate below the fusing point of the aluminium 
fluoride or the aluminous fluorides, and having one element of 
such affinity that it displaces the aluminium in the fluoride 
compound, but having no element capable of combining with 
the reduced aluminium. More specifically, sulphide of silicon 
or an equivalent reagent is mentioned, and the reaction takes 
place on the hearth of a short reverberatory furnace, the 
metallic bath mentioned being preferably metallic iron or cop- 
per, which are able to combine with the resulting aluminium. 

The reaction proposed is so far out of the usual run of specu- 
lations that no opinion can be hazarded as to its probability. 
Further, silicon sulphide is an almost unknown compound, and 
it would be very interesting to know how it is to be prepared ; 
it would probably be a more difficult feat to get the reagent 
alone than to produce aluminium by many well-known methods. 

* U. S. Patent, 410568, Sept. 3, 1889. 


working in aluminium. 

Melting Aluminium. 

DevillE: "To melt aluminium it is necessary to use an 
ordinary earthen crucible and no flux. Fluxes are always use- 
less and almost always harmful. The extraordinary chemical 
properties of the metal are the cause of this ; it attacks very 
actively borax or glass with which one might cover it to pre- 
vent its oxidation. Fortunately this oxidation does not take 
place even at a high temperature. When its surface has been 
skimmed of all impurities it does not tarnish. Aluminium is 
very slow to melt, not only because its specific heat is consid- 
erable, but its latent heat appears very large. It is best to 
make a small fire and then wait patiently till it melts. One can 
very well work with an uncovered crucible." 

" In the fusion of impure aluminium very different phenomena 
are observed, according to the nature of the foreign metal which 
contaminates it. Ferruginous material often leaves a skeleton 
less fusible and pretty rich in iron ; a liquation has taken plafte, 
increasing the purity of the melted material. When the alu- 
minium contains silicon this liquation is no longer possible, or 
at least it is very difficult, and I have sometimes seen some 
commercial aluminium so siliceous that the workmen were 
unable to remelt it. When it is desired to melt pieces together 
they can be united by agitating the crucible or compressing the 
mass with a well-cleaned cylindrical bar of iron. Clippings, 
filings, etc., are melted thus : Separate out first, as far as possi- 
ble, foreign metals, and to avoid their combining with the alu- 
minium heat the divided metal to as low a heat as possible, just 
sufficient to melt it. The oil and organic matters will burn, 



leaving a cinder, which hinders the reunion of the metal if one 
does not pre§s firmly with the iron bar. The metal may then 
be cast very easily, and there is found at the bottom of the 
crucible a little cinder, which still contains a quantity of alu- 
minium in globules. These may be easily separated by rubbing 
in a mortar and then passing through a sieve, which retains the 
flattened globules." 

Biederman gives the following directions : " The whole quan- 
tity of metal which is to be melted must not be put into the 
crucible at once, but little by little, so increasing the mass from 
time to time as the contents become fully melted. The neces- 
sary knack for attaining a good clean melt consists in dipping 
the pieces which are to be melted together in benzine before 
putting in the crucible. Mourey even pours a small quantity 
of benzine into the crucible after the full melting of the metal, 
and he recommends the employment of benzine in the melting 
of all the noble metals. To utilize the scrap pieces produced 
in working aluminium into various useful articles, one must as 
far as possible separate out first the pieces which have been 
soldered, in order that the newly melted aluminium may not be 
contaminated by the solder. The solder adhering to these 
pieces can be removed by treating them with nitric acid, by 
which the aluminium is not attacked." 

Aluminium can be melted with perfect safety in common clay 
or sand crucibles if they are lined with carbon. This can be 
done by mixing lamp-black to a paste with molasses, plaster- 
ing the inside of the crucible evenly and drying slowly for 
several days at a moderate temperature. A means of getting a 
more perfect lining is to ram the crucible full of this paste, dry- 
ing slowly and then hollowing out a cavity of the required size 
leaving a uniform lining of sufficient thickness. In using these 
carbon-lined crucibles they should be kept well-covered, in 
order that the carbon may not burn away too quickly, taking 
particular care to place the cover on when the metal has been 
poured out and the crucible is cooling in the air. Small 
crucibles may be made of a single block of soapstone, and seem 


to last a long while, the aluminium apparently having no action 
on this mineral at a temperature a good deal higher than its 
melting point. 

With extra care, heating in a furnace where the heat is under 
exact control, aluminium may be melted in sand crucibles or in 
cast-iron ones. In using sand or Hessian crucibles no flux 
must be added, and if the metal is heated slowly to a tempera- 
ture only enough above its melting point to admit of pouring 
quickly, it will be found that the crucible is unattacked. If, 
however, the crucible is heated to a bright-red heat at any part, 
it will be found that at that place the aluminium has attacked 
the crucible, and on pouring out the metal a thin, tough skin 
will be left adhering to the spot attacked and generally taking 
with it bits of the wall when it is forcibly detached. This thin 
sheet is hard, tough, and rich in silicon, while the rest of the 
metal poured out has also absorbed a little silicon. An iron 
crucible acts in precisely the same way ; if the heat is kept 
close to the melting point of aluminium the latter does not 
" wet" the iron, but at a bright- red heat it attacks it and ad- 
heres, as in the case of the sarid crucible. I will repeat, that if 
the temperature is kept as low as it is possible to melt and cast 
the metal, and if no fluxes are used, neither the iron nor sand 
crucibles are attacked. 

In the large European works, where 500 or looo lbs. of alu- 
minium are melted at once on the bed of a reverberatory fur- 
nace, the hearth was formerly protected by pure bauxite, 
closely rammed in and strongly fired before using, but basic 
magnesia bricks have now been substituted, similar to those 
used in basic open-hearth furnaces but of purer materials, and 
they are reported as being all that can be desired in their capa- 
city for resisting corrosion. 

A very good lining for any kind of crucible is made by using 
calcined magnesite, for which I recommend the white, Grecian 
mineral, fully shrunk by calcining at the highest heat of a fire- 
brick or porcelain kiln and then ground fine and made into a 
stiff paste with sugar syrup or molasses. The crucible is care- 


fully lined and then heated slowly to redness. On cooling, any 
cracks are carefully filled with more paste, and the crucible is 
baked again ; when properly prepared, however, one baking is 
sufiScient, and the linings do not shrink from the walls of the 
crucible. This lining is a very serviceable one, not only for 
melting aluminium by itself, but also for using flux, for it re- 
sists cryolite, fluorspar or even a siliceous slag perfectly. Since 
experiment has shown that aluminium will take up about 0.25 
per cent, of silicon from one re-melting at a red heat in a best- 
quality graphite crucible, it is very important that such deteri- 
oration be avoided if possible, and I earnestly advise all melt- 
ers and casters of aluminium to take the trouble to prepare 
these magnesite lined crucibles if they wish to get the best re- 
sults. This magnesite can be procured from the importers of 
chemicals in New York city, at a cost of about $0.02 per pound 
for the burnt and ground material. 

Casting Aluminium. 

Deville: "Aluminium can be cast very easily in metallic 
moulds, but better in sand for comphcated objects. The 
mould ought to be very dry, made of a porous sand, and should 
allow free exit to the air expelled by the metal, which is vis- 
cous when melted. The number of vents ought to be very 
large, and a long, perfectly round git should be provided. 
The aluminium, heated to redness, ought to be poured rather 
quickly, letting a little melted metal remain in the git till it is 
full, to provide for the contraction of the metal as it solidifies. 
In general, this precaution ought to be taken even when alu- 
minium is cast in iron ingot moulds or moulds of any other 
metal. The closed ingot moulds give the best metal for roll- 
ing or hammering. By following these precautions, castings 
of great beauty may be obtained, but it is not advisable to con- 
ceal the fact that to be able to succeed completely in all these 
various operations requires for aluminium, as for all other 
metals, a special familiarity with the material which practice 
alone is able to give." 


The peculiarity of molten aluminium which a metal caster 
would first notice is its viscousness, that is, it runs thick. 
When about to pour, a thick edge or lip of metal forms, which 
must in many cases be punctured in order to start the metal 
flowing. On account of this property it does not run sharply 
in the moulds except where a head of metal puts it under pres- 
sure. To obtain sharp castings there must be a " gate" to give 
pressure to the metal, and when we remember that the gravity 
of aluminium is only 2.6 and that, besides, it does not flow 
thinly, it will be seen that a much higher head of metal is 
necessary to ensure sharp castings than is needed for iron, 
brass, etc. A small slab of» aluminium two inches high run in 
a closed iron mould with very little gate was quite sharp at the 
lower end but had rounded corners at the top. For casting in 
closed moulds, the best results as to sharp castings free from 
cavities are obtained if a slight artificial pressure can be ap- 
plied to the still liquid metal in the mould immediately after 
pouring. This is accomplished by Dr. C. C. Carroll, of New 
York, an expert in making cast-aluminium dental plates, by 
closing in tightly the top of the crucible containing the molten 
aluminium, and then by air pressure from a rubber bulb forcing 
the metal through a syphon-shaped tube terminating under- 
neath the metal and connecting tightly with the pouring gate of 
the mould. The mould is previously heated, in order not to 
chill the metal too quickly, and when the metal has been forcfed 
out of the crucible by squeezing the bulb the pressure is con- 
tinued for a short time in order to force the metal into every 
crevice of the mould and allow the casting to set under pres- 

The idea is undoubtedly correct, and the excellently sound 
and extremely sharp castings obtained by Dr. Carroll attest its 
success in practice. 

The Passaic Art Casting Company, of Passaic, N. J., make 
very fine castings in aluminium by the Smith pressure-casting 
process, by which the finest details of engraved, chased or re- 
poussee work are brought out with a finish equal to electro- 


types. Moulds, either of sand or of a liquid plaster-of-paris 
composition, are made in as flat a shape as possible so that they 
may be piled on top of each other, the gates from each mould 
leading to a central feeder. The entire charge of molten metal 
is kept in a suitable holder above the moulds. The moulds 
being air-tight, an air-pump is connected with the bottom of the 
molds while pressure is put on the molten metal above. A 
thin diaphragm used to keep the molten metal in the receiver 
is burst by the pressure, allowing the aluminium to flow quickly 
into the moulds, where it fills every minute cavity. Extremely 
light and sharp castings, having remarkable solidity and free 
from blow-holes, are thus obtained. 

Aluminium can be cast in dry sand moulds, which are best 
lined with powdered plumbago. The metal should be poured 
fast, and not be far above its melting point, or the castings will 
not be sound. The shrinkage to be allowed for is very nearly 
one-quarter of an inch to the foot. 

Quite recently a large amount of cast aluminium hollow-ware 
has been made for culinary use. Tea-kettles have been cast in 
one piece with walls only one-sixteenth of an inch thick. These 
castings possess the advantage that they are more malleable 
than any other cast metal. Thin castings of tea-kettles have 
been hammered in and bent almost double before breaking, 

Mr. W. S. Cooper, of Philadelphia, makes some remarkably 
large sand-castings, adding a little copper to reduce the shrink- 
age. Full-sized bath-tubs, 6 feet long, 2 feet wide, 2 feet deep, 
and three-eighths inch thick, with a roll rim, and weighing com- 
plete 140 lbs., are worthy of special mention, for such a cast- 
ing would not be easy to make in the best-flowing metal. 

Purification of Aluminium. 

Freeing from slag. — Deville gives the following information 
on this important subject : 

" It is of great importance not to sell any aluminium except 
that which is entirely free from the slag with which it was pro- 
duced and with which its whole mass may become impregnated. 


We have tried all sorts of ways of attaining this end, so as to ob- 
tain a metal which would not give any fluorides or chlorides 
upon boiling with water, or give a solution which would be pre- 
cipitated by silver nitrate. At Glaciere we granulated the metal 
by pouring it while in good fusion into water acidulated with 
sulphuric acid ; this method partially succeeded. But the pro- 
cess which M. Paul Morin uses at present (1859), and which 
seems to give the best results, is yet simpler. Three or four 
kilos of aluminium are melted in a plumbago crucible without 
a lid, and kept a long time red-hot in contact with the air. Al- 
most always acid fumes exhale from the surface, indicating the 
decomposition by air or moisture of the saline matter impreg- 
nating the metal. The crucible being withdrawn from the fire, 
a skimmer is put into the metal. This skimmer is of cast iron ; 
its surface ought not to be rough and it will not be wetted by 
the aluminium in the least during the skimming; it may be of 
advantage to oxidize its surface with nitre before using. The 
white and slaggy matters are then removed, carrying' away also 
a little metal, and are put aside to be remelted. So, in this 
purification, there is really no loss of metal. After having 
thus been skimmed, the aluminium is cast into ingots. This 
operation is repeated three or four times until the metal is per- 
fectly clean, which is, however, not easily told by its appearance ; 
for, after the first fusion, the crude aluminium when cast into 
ingots has a brilliancy and color such as one would judge quite 
irreproachable, but the metal would not be clean when it was 
worked, and especially when polished would present a multi- 
tude of little points called technically ' piqiires,' which give to 
its surface, especially with time, a disagreeable look. Alumin- 
ium, pure and free from slag, improves in color on using. It 
is the contrary with the impure metal or with aluminium not 
free from slag. When aluminium is submitted to a slow, corrod- 
ing action, its surface will cover itself uniformly with a white, 
thin coating of alumina. However, any time that this layer is 
black or the aluminium tarnishes, we may be sure that it con- 
tains a foreign metal, and that the alteration is due to this im- 


Freeing from impurities. — Again Deville is the authority, and 
we quote his advice on the subject: — 

" A particular characteristic of the metallurgy of aluminium 
is that it is necessary, in order to get pure metal, to obtain it so 
at the first attempt. When it contains silicon, I know of no 
way to eliminate it: all the experiments which I have made on 
the subject have had a negative result; simple fusion of the 
metal in a crucible, permitting the separation by liquation of 
metals more dense, seems rather to increase the amount of sil- 
icon than to decrease it. When the aluminium contains iron 
or copper, each fusion purifies it up to a certain limit, and if 
the operation is done at a low heat there is found at the bottom 
of the crucible a metallic skeleton containing much more iron 
and copper than the primitive alloy. At first I made this liqua- 
tion in the muffle of a cupel furnace, in which process the ac- 
cess of air permitted the partial oxidation of these two metals. 
The little lead which aluminium may sometimes take up may 
thus be easily separated. Unfortunately, the process does not 
give completely satisfactory results. It is the same in fusing 
impure aluminium under a layer of potassium sulphide, KjSj; 
there is a partial separation of the lead, copper, and iron. That 
which has succeeded best with us is the process which we have 
employed for a long time at Glaci^re, and which consists in 
melting the aluminium under nitre in an iron crucible. We 
have in this way improved the quality of large quantities of 
aluminium. The operation is conducted as follows : Aluminium 
has generally been melted with nitre in order to purify it by 
means of the strong disengagement of oxygen at a red heat, 
no doubts being entertained as to the certainty of the result. 
But it is necessary to take great care when doing this in an 
earthen crucible. The silica of the crucible is dissolved by the 
nitre, the glass thus formed is decomposed by the aluminium, 
and the siliceous aluminium thus formed is, as we know, very 
oxidizable, and especially in the presence of alkalies. So, the 
purification of aluminium by nitre ought to be done in a cast- 
iron crucible well oxidized itself by nitre on the inside." 


" On mfelting aluminium containing zinc in contact with the 
air and at a temperature which will volatilize the zinc, the 
largest part of the latter burns and disappears as flaky oxide. 
To obtain a complete separation of the two metals it is neces- 
sary to heat the alloy to a high temperature in a brasqued cru- 
cible. This experiment succeeds very well, but it is here shown 
that the aluminium must oxidize slightly on its surface, for 
some carbon is reduced by the aluminium from the carbonic 
oxide with which the crucible is filled. This carbon thus sep- 
arated is quite amorphous.'' 

Dr. Lisle, of Springfield, O., tested the removal of zinc from 
aluminium by distillation and succeeded in obtaining a product 
with 98 per cent, of aluminium from which the zinc had been 
completely removed. If aluminium containing tin is melted 
with lead, the latter sinks with the tin, removing it almost com- 
pletely from the aluminium, but the metal remaining retains a 
little lead. In an experiment made by the author, aluminium 
with 10 per cent, of tin was treated in this way, but the alu- 
minium retained nearly 7 per cent, of lead, giving it a blue 
color and large crystalline structure. M. Peligot is reported by 
Deville to have succeeded in cupelling aluminium with lead, 
whereby impure, tough metal became quite malleable. It may 
be that the small percentage retained by aluminium when 
melted with lead is removed by fusion on a cupel, but I have 
been unable to perform any operation with aluminium at all 
analogous to the cupellation of silver with lead. 

G. Buchner states that commercial aluminium contains con- 
siderable quantities of silicon, which by treatment, when melted, 
with hydrogen, evolves hydrogen silicide. This does not result 
if arsenic is present. 

To test this point, I took a sample of aluminium containing 4 
per cent, of silicon, and 94 per cent, of aluminium. This was 
melted at a low red heat in a sand crucible and hydrogen gas 
run in for 12 minutes. On pouring out, the crucible was unat- 
tacked, but the metal was identical in color and structure with 
that not treated. This was repeated several times, hydrogen 


gas being passed in as long as 20 minutes with the metal at red 
heat, but no apparent change in the purity of the metal re- 

Similar experiments were made with a current of sulphuretted 
hydrogen gas, with a view of removing the iron. No improve- 
ment in the looks of the aluminium was apparent. If, however, 
air was blown into the melted metal, an improvement was made. 
The aluminium was at a bright red heat, and air blown in for 5 
minutes, at the end of which time about 5 to 10 per cent, of 
dross composed of mixed metal and oxide was formed, and the 
remaining metal was whiter, of a finer grain, and evidently much 

Prof. Mallet * made some very accurate estimations of the 
atomic weight of aluminium (which he found to be 27.02), and 
obtained the chemically pure metal required for his work by 
the following process, which is quite applicable when studying 
the properties of the pure metal, yet is of course altogether out 
of the question as an industrial operation. The purest com- 
mercial metal was bought, and on analysis was found to con- 
tain — 

Aluminium 96.89 per cent. 

Iron 1.84 " 

Silicon 1.27 " 

This was treated with liquid bromine and converted into bro- 
mide. This salt was then purified by fractional distillation, the 
temperature being very carefully regulated, and the operation 
repeated until the product was perfectly colorless, and dissolved 
in water without leaving any perceptible impurity. The reduc- 
tion was accomplished with difiSculty and much loss by treat- 
ing with sodium in a crucible made of a mixture of pure alu- 
mina and sodium aluminate. The metal obtained gave on 
analysis no weighable quantity of impurities, and the properties 
mentioned in Chapters III. and IV. as Mallet's observations are 
those quoted from his report on this pure metal. 

* Philosophical Magazine, 1880. 


Professor Le Verrier states * that by a simple fusion of alu- 
minium under a layer of an alkaline fluoride the amount of sili- 
con is easily diminished. He states that a specimen containing 
0.81 per cent, of silicon, contained after one such fusion only 
0.57 per cent. 

If aluminium is overheated in the crucible, it has a great 
tendency to absorb gas (like silver), and gives it out in setting, 
producing blow-holes in the castings. It is therefore important 
to carefully regulate the temperature of melting. Coehn re- 
commends the use of a very small amount of sodium, to take 
gas out of the metal, only enough being used to form a thin 
film over the melted metal. 


By heating to very low redness and cooling quickly, as by 
dropping into water, aluminium becomes soft. To get the best 
results, the metal should be heated until it just begins to glow 
in the dark, or the object is rubbed with a lump of fat, and the 
moment that the black trace left by the carbonization of the fat 
disappears, the metal is removed from the annealing oven. 
Great care is necessary in annealing thin sheets which are being 
beaten into leaf, to avoid melting them. Fine wire can be an- 
nealed over the chimney of an Argand burner. If it is neces- 
sary that the heat must be kept below the point where the 
metal would bend and lose its shape, the articles can be heated 
in boiling linseed oil and allowed to cool very gradually with 
the oil. Very thin sheets or wire can be softened by putting 
into boiling water and letting them cool with the water. (See 
also pp. 70, 74.) 


By hammering, rolling, or drawing, aluminium becomes sen- 
sibly harder and stiffer, so that it becomes elastic enough to be 
used even for hair-springs for watches. If castings of alumin- 
ium are made a little larger than the finished article desired, 

♦ Comptes Rendus, July, 1894, p. 13. 


and drop-forged cold in dies of the right size, the hardness is 
much increased, and it becomes much better able to stand wear 
in bearings where great weight does not have to be endured, 
e.g., in surveying instruments. Queen &Co., of Philadelphia, 
state that they find it best, however, to bush all sockets in alu- 
minium castings with brass, when making a bearing or tapping 
in a screw. 

The most careful tests fail to show the slightest increase in 
hardness in metal which has been chilled, as practised with 
cast-iron. Surfaces cast in cold, metallic moulds seem as soft 
as those cast in sand. 

The Eureka Tempered Copper Company of Northeast, Penn- 
sylvania, claim that they have hardened aluminium castings in 
the same way as they harden copper, but I have seen none 
of their work. 


Before rolling a bar of aluminium it is well to soften it, and 
to taper down a "lead" by hammering. The metal forges well 
under the hammer at a low red heat. The metal for rolling had 
better be cast into plates in covered iron ingot-moulds, and the 
surfaces planed to remove small irregularities. The rolling is 
not difficult, except that a large amount of power is required 
(about as much for cold aluminium as for hot steel), and as the 
metal quickly gets hard it must be annealed often. It is 
recommended that the metal be brought warm under the rolls, 
and, if possible, elongated lo to 20 per cent, with the first pass, 
in order to entirely destroy the crystalline structure of the 
metal. The annealing is repeated between each pass until the 
sheet is about 3 millimetres thick, after which it can generally 
be rolled with fewer annealings, sometimes without any at all. 
Rolls warmed to 100-150° C. work better than cold ones. Alu- 
minium has been rolled down to the thinness of tissue-paper. 

Thin-rolled sheets may be still further extended out by beat- 
ing into leaf. The gold-beaters are said to have no special 
difficulty in doing this, except that more frequent annealings 


are necessary than with gold or silver. M. Degousse was the 
first to make this leaf, in 1859; he states that the tempering 
must be done by warming only to 100° or 150° C, an actual 
glowing heat proving very unsuitable. Aluminium leaf is made 
as thin as ordinary gold or silver leaf, and this property would, 
therefore, establish aluminium as next to these noble metals in 
the order of malleability. 

The Mannesmann process of cold rolling is applicable to alu- 
minium, by which tubes of all shapes and sizes can be made. 

Mr. Elwood Ivins, of Philadelphia, has a special process for 
making metallic tubes, which is a sort of rolling process on a 
mandril, by which he makes aluminium tubes of any size, shape 
or thickness of walls. I have seen a tube made by this process 
10 metres long by i millimetre in diameter. 


Prof. Thurston places aluminium as sixth in the order of duc- 
tility, being preceded by gold, silver, platinum, iron and cop- 
per, but it is doublful if it does not rank equally as high as 
iron. Deville states that in 1855 M. Vangeois obtained very 
fine wire with metal far from being pure. Bell Bros., at New- 
castle-on-Tyne, recommended that the metal for drawing be run 
into an open mould so as to form a flat bar of about one-half 
inch section, the edges of which are beaten very regularly with 
a hammer. The diameter should be very gradually reduced at 
first, with frequent heating. When the threads are required 
very fine the heating becomes a very delicate operation, on ac- 
count of the fineness of the threads and the fusibility of the 
metal. The gauge should be reduced by the smallest possible 
gradations, when wires may be obtained as fine as a hair. 

Aluminium tubes are drawn, either round or square, from 
sheet which has been soldered together or from cast rings of 
the required section ; the first method is said to be preferred. 
As the aluminium quickly loses its temper it must be annealed 
after each extension. 

working in aluminium. 455 

Stamping and Spinning. 

Aluminium can be spun on the lathe into all sorts of round 
and hollow forms ; it may also be pressed or stamped into shape 
in the cold ; it is of* advantage in doing this to use a kind of 
varnish composed of 4 parts of oil of turpentine and i part of 
stearic acid. Soap-water is always to be avoided. 

Grinding, Polishing and Burnishing. 

When cast carefully it can be filed without fouling the file. 
Spun and stamped articles of aluminium can easily be ground 
by using olive oil and pumice. A fine steel scratch-brush run 
at a high speed will give a good polish to sand castings, and 
remove the yellowish streaks that may have been produced by 
too hot metal. If sheet metal is thus treated it gives a frosted 
appearance, which forms a beautiful finish. Biederman makes 
the following remarks on polishing : " The use of the old means 
of polishing and burnishing metals, such as soap, wine, vinegar, 
linseed-oil, decoction of marshmallow, etc., is not effective with 
aluminium, but, on the contrary, is even harmful, because, 
using them, the blood stone and the burnishing iron tear the 
metal as fine stone does glass. Oil of turpentine has also been 
used, but with no good effect. Mourey found, after many at- 
tempts, that a mixture of equal weights of olive oil and rum, 
which were shaken in a bottle till an emulsified mass resulted, 
gave a very brilliant polish. The polishing stone is dipped in 
this liquid, and the metal polished like silver, except that one 
must not press so hard in shining up. The peculiar black 
streaks which form under the polishing stone need cause no 
trouble ; they do not injure the polish in the least, and can be 
removed from time to time by wiping with a lump of cotton. 
The best way to clean a soiled surface and remove grease is to 
dip the object in benzine, and dry it in fine sawdust." 

Mr. J. Richards found that when buffing in the ordinary way 
the dark-colored burnishing powder cut into the metal and 
filled its pores with black specks. He found the best means of 


burnishing is to use a piece of soft wood soaked in olive oil ; 
this closes the grain, and gives a most brilliant finish. 

The Pittsburgh Reduction Company polish sheet aluminium 
with a sheep-skin, chamois-skin or rag buff, using rouge, the 
same as for brass. The particular requisite of a polishing pow- 
der for aluminium is that it be very finely ground. This com- 
pany sells a polishing powder called "Almeta Polish," which 
consists of 

Stearic acid i part. 

Fuller's earth i " 

Rotten stone 6 " 

The whole ground very fine and well mixed. 


Kerl & Stohman : Aluminium resists the action of the en- 
graving tool, which slides upon the surface of the metal as 
upon hard glass. But as soon as a varnish of 4 parts of oil of 
turpentine and i of stearic acid, or some olive oil mixed with 
rum is used, the tool cuts into it as into pure copper. 

Cleaning and Pickling: Mat. 

Dirt and grease are removed easily by dipping in benzine or 

To obtain the beautiful "mat" or frosted efifect, which alu- 
minium takes as easily as silver, Deville recommended dipping 
for an instant in a very dilute solution of caustic soda, washing 
in a large quantity of water, and last dipping in strong nitric 
acid. An English writer recommends dipping i^ minutes in 
a solution of 3 ounces of caustic potash to a quart of water, 
wash thoroughly, and then dip in a mixture of 3 measures 
strong nitric acid to 2 measures sulphuric acid. By the above 
treatments, any elements which might contaminate the alumin- 
ium, except a very large proportion of silicon, are dissolved 
from the surface. If, as Mourey recommends, some hydroflu- 
oric acid is added to the nitric, the silicon is more actively 

working in aluminium. 45 7 


Aluminium cannot be welded except by the electrical pro- 
cess, and then only with difficulty. The high electric conduc- 
tivity, the high specific heat and latent heat of fusion, and par- 
ticularly the mushy state into which the metal passes some 
considerable time before its real melting point, make the oper- 
ation hard to perform satisfactorily. In the early days of elec- 
tric welding it could not be done at all, but the Thomson Elec- 
tric Welding Company state that they have constructed a 
special machine for this kind of material, and can now make 
the welds rapidly and satisfactorily. 

Soldering Aluminium. 

A satisfactory solder for use on any metal should fill the fol- 
lowing requirements :* 

1. It should fuse readily. 

2. It must alloy easily with the metal; in common parlance, 
it must bite. 

3. It must be tough. 

4. It must not disintegrate. 

5. It should be of the same color as the metal. 

6. It should not discolor with age. 

7. It should not be too expensive. 

Aluminium is a difficult metal to solder properly for three 
particular reasons : 

I. It is difficult to expose a bare surface of aluminium to the ^ 
action of the solder. Grease and dirt can be easily enough 
removed, but there is always present a thin, continuous coating 
of alumina which effectually prevents the solder from getting 
to the metal beneath. It might be thought that a thorough 
sand-papering or fiHng of the surfaces immediately before sol- 
dering would overcome this, but we find it of absolutely no 
use ; the reason is that the thin, invisible skin of alumina forms 

* For a more extended essay on the properties of a good solder, see a paper by the 
writer in the "Aluminium World," New York, October, 1894. 


instantly, as quickly as the metal is laid bare. The ordinary 
fluxes used in soldering have no effect at all on this coating. 
To overcome this difficulty, it is necessary either to use some 
particular flux which can remove this oxide and preserve the 
surface clean until the solder can take hold, or else to incor- 
porate into the composition of the solder some ingredient 
which can do the same thing. The latter device has been the 
most successful, as theoretically it should be ; for if the solder 
as it were carries its own flux, the action of the flux and the 
taking hold of the solder must be simultaneous, and therefore 
most effective. 

2. Aluminium has a high conductivity for heat. This inter- 
feres greatly with the attempt to produce a local temperature 
at the joint high enough for the solder to alloy ; which causes 
the solder to chill quickly in the joint, particularly when work- 
ing on bulky articles. The only way to overcome this diffi- 
culty is to use very hot soldering-bolts and to work slowly, so 
as to get the metal at the joints hot. 

3. Aluminium is highly electro-negative. This property 
causes the first difficulty, the instantaneous formation of a thin 
skin of oxide, but it also operates in another way after the joint 
is made. It causes galvanic action to arise at the joint between 
the aluminium and the solder, and causes the joint to weaken 
rapidly. To render this action as small as possible, the solder 
should be made as far as practicable of metals near to alu- 
minium in the galvanic series, such particularly as zinc. 

With the preceding introduction to the subject, we will con- 
sider the various solders which have been proposed and used, 
giving first Deville's views on the question, for even at the time 
of writing his book (1859), the difficulty of soldering alumin- 
ium properly was regarded as one of the greatest obstacles to 
its industrial employment. 

Deville : " Aluminium may be soldered, but in a very im- 
perfect manner, either by means of zinc, or cadmium, or alloys 
of aluminium with these metals. But a very peculiar difficulty 
arises here — we know no flux to clean the aluminium which 


does not attack the solder, or which, protecting the solder, 
does not attack the aluminium. There is also an obstacle in 
the particular resistance of aluminium to being wetted by the 
more fusible metals, and on this account the solder does not 
run between and attach itself to the surfaces to be united. M. 
Christofle and M. Charriere made, in 1855, during the Exposi- 
tion, solderings with zinc or tin. But this is a weak solder and 
does not make a firm seam. MM. Tissier, after some experi- 
ments made in my laboratory, proposed alloys of aluminium 
and zinc, which did not succeed any better. However, M. 
Denis, of Nancy, has remarked that whenever the aluminium 
and the solder melted on its surface are touched by a piece of 
zinc, the adhesion becomes manifest very rapidly, as if a partic- 
ular electrical state was determined at the moment of contact. 
But even this produces only weak solderings, insufficient in 
most cases." 

" A long time ago, M. Hulot proposed to avoid the difficulty 
by previously covering the piece with copper, then soldering 
the copper surfaces. To efifect this, plunge the article, or at 
least the part to be soldered, into a bath of acid sulphate of 
copper. Put the positive pole of a battery in communication 
with the bath, and with the negative pole touch the places to 
be covered, and the copper is deposited very regularly. M. 
Mourey has succeeded in soldering aluminium by processes yet 
unknown to me ; samples which I have seen looked excellent. 
I hope, then, that this problem has found, thanks to his ingenu- 
ity, a solution ; a very important step in enlarging the employ- 
ment of aluminium." 

Mourey's first practicable solders for aluminium were of zinc 

and aluminium, and of two kinds — hard and soft. He used a 

soft solder to first unite the pieces, and afterwards finished the 

soldering with a less fusible one. These solders contained — 

I. II. 

Aluminium 20 15 

Zinc 80 85 

The alloys with the larger proportion of zinc are the easiest 











melting or softest ones, one such as IV. being used for ordin- 
ary work, while II. was used for brazing. These solders have 
the disadvantage that on melting they oxidize very easily, and 
in consequence of the film of oxide thus formed the work is so 
much more difficult. This difficulty can be overcome by dip- 
ping the small grains of solder in copaiva balsam and turpen- 
tine, which keep out the air, and act as reducing agents during 
the operation. The new solders subsequently used were sim- 
pler in application, for the work was finished with one solder 
and the moistening with balsam was rendered unnecessary. 

Mourey improved upon the zinc-aluminium solders by add- 
ing copper, using five different alloys of these three metals ac- 
cording to the objects to be soldered. They contained — 

I. II. III. IV. v. 

Aluminium 12 9 7 6 4 

Copper 8 6 5 4 2 

Zinc 80 85 88 90 94 

The following directions are given for preparing these alloys:* 
" To make the solder, first put the copper in the crucible. 
When -it is melted, then add the aluminium in three or four 
portions, thereby somewhat cooling the melted mass. When 
both metals are melted, the mass is stirred with a small iron 
rod, and then the required quantity of zinc added, free from 
iron, and as clean as possible. It melts very rapidly. The 
alloy is then stirred briskly with an iron rod for a time, some 
fat or benzine being meanwhile put in the crucible to prevent 
contact of the metal with air and oxidation of the zinc. Finally 
the whole is poured out into an ingot mould previously rubbed 
with benzine. After the addition of zinc, the operation must 
be finished very rapidly, because the latter will volatilize and 
burn out. As soon as the zinc is melted the crucible is taken 
out of the fire. Only zinc free from iron can be used, since 
even an apparently insignificant amount of this impurity injures 
the qualities of the solders very materially in regard to dura- 
bility and fusibility. 

* Das LSthen, by Edmund Schlosser, \ ienna, 1880, p. loi. 


" The separate pieces of metal to be soldered together are 
first well cleaned, then made somewhat rough with a file at the 
place of juncture, and the appropriate solder put on it in pieces 
about the size of millet grains. The objects are laid on some 
hot charcoal, and the melting of the solder effected by a blast 
lamp or a Rochemont turpentine-oil lamp. During the melt- 
ing of the solder, it is rubbed with a little soldering iron of 
pure aluminium. The soldering iron of pure aluminium is 
essentially a necessity for the success of the operation, since an 
iron of any other metal will alloy with the metals composing 
the solder, while the melted solder does not stick to the iron 
made of aluminium. 

" For quite small objects, as for jewelry, solder I. is used ; for 
larger objects and ordinary work IV. is more suitable, and is the 
solder most used. These alloys work so perfectly that plates 
soldered together never break at the joint when bent back and 
forth, but always give way in other places ; which is a result 
not always possible in the best soldering of plates of silver." 

Bell Bros, used the above solders in their works at Newcastle, 
and in a description of the soldering operation state the follow- 
ing facts in addition to those already given :* " In the operation 
of soldering, small tools of aluminium are used, which facilitate 
at the same time the fusion of the solder and its adhesion to 
the previously prepared surfaces. Tools of copper or brass 
must be strictly avoided, as they would form colored alloys 
with the aluminium and the solder. The use of the little tools 
of aluminium is an art which the workman must acquire by 
practice. At the moment of fusion the work needs the appH- 
cation of friction, as the solder suddenly melts very completely. 
In soldering it is well to have both hands free, and to use only 
the foot for the blowing apparatus." 

We also find the following alloys credited to Mourey as used 
by him for soldering aluminium, probably with the blow-pipe :t 

* Chemical News, iv., 81. 
t Dingier, 166, 205. 


Aluminium 30 20 

Copper 20 15 

Zinc 50 65 

While it is true that Mourey received from the Societe 
d'Encouragement a prize for the solders he had found, and 
while they were the best then in use, yet every one knows that 
they were not satisfactory. Their melting points were too high, 
and it required great skill to make neat joints. Even as late as 
1890 we find a writer saying, " The want of a suitable solder for 
aluminium has greatly retarded its use in the arts," although 
many attempts had meanwhile been made to provide solders 
superior to Mourey's. 

Col. Wm. Frishmuth * recommends a solder containing: — 

Aluminium 20 

Copper 10 

Zinc 30 

Tin 60 

Silver 10 

Later, Col. Frishmuth f states that the solder just given is 
used for fine ornamental work, while for lower-grade work he 
uses the following : — 


Bismuth . 









' 2-1 

He recommends for a flux, in all cases, either paraffin, stearin, 
vaselin, copaiva balsam, or benzine. In the solder for fine 
work, if aluminium is used in larger quantity than recom- 
mended, the solder becomes brittle. 

Schlosser % recommends two solders containing aluminium as 
especially suitable for soldering dental work, on accbunt of 
their resistance to chemical action. Copper cannot be allowed 
in alloys intended for this use, or only in very insignificant 

* Techniker, vi. 249. 

t Wagner's Jahresb., 1884. 

X Das Lothen, p. 103. 


quantity, since it is so easily attacked by acid food, etc. Since 
these two alloys can probably be used also for aluminium dental 
work, we subjoin their composition — 

Platinum-aluminium solder. Gold-aluminium solder. 

Gold 30 Gold.' 50 

Platinum I Silver 10 

Silver 20 Copper 10 

Aluminium 100 Aluminium 20 

M. Bourbouze states that the difficulties met with in solder- 
ing aluminium are satisfactorily overcome by the following 
process : — 

*"The parts to be united are subjected to the ordinary op- 
eration of tinning, except that in place of pure tin an alloy of 
tin and zinc, or, better, of tin, bismuth and aluminium is used. 
However, preference is given to an alloy of tin and aluminium, 
mixed in different proportions to suit the work put on the joint. 
For those which are to be subsequently worked, an alloy of 45 
parts tin and 10 parts aluminium should be used. This solder 
is malleable enough to resist hammering, drawing, or turni